CN106636616A - Sintered ore yield prediction method based on bellows waste gas temperature - Google Patents

Abstract

The invention discloses a sintered ore yield prediction method based on a bellows waste gas temperature. The method comprises the steps of: determining heat state parameters influencing on a sintered ore yield; building a history production data sample database of sintering parameters; calculating the sintered ore yield; adopting a polynomial fitting method to perform the curve fitting for bellows waste gas temperature data, and obtaining numerical values of the heat state parameters through a differential extremum method; building a sintered ore yield prediction model based on the numerical values of the heat state parameters; performing the curve fitting for bellows waste gas temperature data of an ore to be predicted to obtain numerical values of the heat state parameters of the ore to be predicted; and inputting the numerical values of the heat state parameters of the ore to be predicted into the sintered ore yield prediction model, and outputting a variable as the yield of the ore to be predicted. The method can accurately predict the sintered ore yield, and adjusts process parameters in real time in the sintering process to increase the sintered ore yield and to provide important basis for energy conservation and emission reduction.

Description

A kind of sintering deposit yield prediction method based on bellows EGT

Technical field

It is the present invention relates to steel sintering process produces energy-saving field more particularly to a kind of based on bellows EGT Sintering deposit yield prediction method.

Background technology

Steel and iron industry is one of pillar industry in national economy, and the development of steel and iron industry will also determine sending out for Chinese national economy Exhibition.Iron and steel is widely used in the industry such as national defence, traffic, building, machine-building, automobile, has very in the national economic development Important strategic position.Iron ore powder sintering is one of important step in steel manufacture process, is to ensure that blast furnace obtains high-quality and burns The key point of knot ore deposit, it adds a certain amount of fuel and solvent in ore to iron content, tile after mixing granulation Sintering ignition is carried out on sintering machine, makes compound that a series of physical and chemical reaction to occur under the high temperature conditions, generation High-quality is containing the primary raw material that iron agglomerate is blast fumance.In iron and steel production, the quality of sintering deposit and yield effect iron and steel Quality and yield, govern the growth of business economic productivity effect.The Forecasting Methodology of sintering deposit yield rate is conducive to look-ahead The yield of sintering deposit, so as to realize adjusting raw material parameter, device parameter and sintering operation parameter in advance, for raising sintering deposit Quality and yield have important effect.

Iron and steel production process is more, technological process is long, and sintering process mainly includes：Sintered material, mixing granulation, segregation cloth The process procedures such as material, ignition, the broken, cooling screening of hot ore deposit.At present, the sintering machine used in sintering process is typically all Strand exhaust sintering machine, is made up of material-feeding mechanism, main frame, igniter, large flue, Water-seal zipper conveyor etc..The technique of sintering process Flow process is shown in accompanying drawing 1.

In sintering production process, the hot environment (Warm status) that fuel combustion is provided in compound is to affect sinter quality With the most important procedure parameter of yield, it is to predict that the quality and yield of sintering deposit are crucial that it is accurately identified.In sintering process The height of compound local environment temperature, the length of hot environment retention time and compound complete the position residing for sintering process Put, can all affect the yield rate and quality of sintering deposit.The lamination of sinter bed is shown in accompanying drawing 2.

At present, workman is mainly predicted sintering deposit finished product by sintering end point temperature and position in sintering production process Rate, fuel amount of allocating and sintering machine speed are adjusted according to testing result, but this detection mode is often burnt in compound Just can carry out after the completion of knot, with certain hysteresis quality, while only according to the temperature information at sintering end point moment, it is impossible to enough complete Warm status of the reaction compound in face ground in whole sintering process；And this prediction is mainly by controlling sintering end point temperature Be in certain scope with position, so as to judge the quality of sintering deposit and the situation of yield, not directly to sintering deposit into Product rate is predicted.Therefore, the yield rate that sintering deposit is directly predicted by sintering process Warm status has weight to blast furnace ironmaking The meaning wanted.

The content of the invention

In view of this, The embodiment provides a kind of can be carried out accurately in advance to the yield rate of sintering process sintering deposit The sintering deposit yield prediction method based on bellows EGT surveyed.

Embodiments of the invention provide a kind of sintering deposit yield prediction method based on bellows EGT, including following Step：

(1) the Warm status parameter of the yield rate for affecting sintering deposit is determined according to the sintering process of sintering deposit；

(2) historical production data to affecting the sintering parameter of sintering process is carried out at zero-phase filtering and sequential registration Reason, and sampling processing is carried out to the historical production data, set up historical production data sample database；

(3) yield rate of sintering deposit is calculated using the historical production data sample database；

(4) the bellows EGT data in the historical production data sample database are entered using polynomial fitting method Row curve matching obtains a fitting function, and by carrying out differential to the fitting function extreme value is sought, and obtains Warm status parameter Numerical value；

(5) numerical value of the Warm status parameter obtained using step (4) is used as input variable, with the sintering that step (3) is obtained The yield rate of ore deposit is computed repeatedly as output variable and verified, sets up sintering deposit yield prediction model；

(6) the bellows EGT data of ore deposit to be predicted are carried out curve fitting and by differential using polynomial fitting method Extremum method is asked to obtain the numerical value of the Warm status parameter of ore deposit to be predicted；

(7) numerical value of the Warm status parameter of the ore deposit to be predicted for obtaining step (6) as input variable be input into sintering deposit into Product rate forecast model, the output variable of the sintering deposit yield prediction model is the yield rate of ore deposit to be predicted.

Further, in the step (1), Warm status parameter is obtained by analyzing the mechanism of sintering process, and analysis determines Affecting the Warm status parameter of the yield rate of sintering deposit includes bellows high-temperature temperature, high temperature hold time, sintering end point temperature and burning Knot final position.

Further, the sintering parameter includes bellows EGT, machine speed, little Cheng Kuang, returns mine and great achievement ore deposit.

Further, in the step (2), the selected sampling period carries out sampling processing to historical production data, described to adopt The sample cycle is the period of waves of great achievement ore deposit, and the period of waves of the great achievement ore deposit is 45min.

Further, after the yield rate of sintering deposit refers to that sintering process terminates, finished product sintering deposit accounts for the proportion of sinter cake, institute The computing formula for stating the yield rate of sintering deposit is as follows：

In formula：ρ represents the yield rate (%) of sintering deposit, Q_DRepresent great achievement mineral products amount (Kg/h) of sintering, Q_XRepresent sintering It is little into mineral products amount (Kg/h), Q_FRepresent the quantity of return mines (Kg/h) of sintering；It is the great achievement mineral products amount, little into mineral products amount and quantity of return mines Data are obtained from historical production data sample database.

Further, the numerical value for obtaining Warm status parameter is comprised the following steps：

(4.1) bellows EGT data sample is chosen from the historical production data sample database, if bellows number Mesh is M, uses (X_i,T(X_i)) represent a sample data, i=1,2 ... M, X_iWith a distance from representing i-th bellows away from igniting, T (X_i) represent with a distance from away from igniting to be X_iBellows exhaust gas temperature value；

(4.2) polynomial fitting method is adopted, to one group of sample data (X in bellows EGT data sample_i,T(X_i)) Fitting of a polynomial is carried out, fitting function is obtained for T (X_i)=a₈X_i ⁸+a₇X_i ⁷+a₆X_i ⁶+a₅X_i ⁵+a₄X_i ⁴+a₃X_i ³+a₂X_i ²+a₁X_i ¹+ a₀,

a₈、a₇、a₆、a₅、a₄、a₃、a₂、a₁And a₀Respectively it is fitted the coefficient for obtaining；

(4.3) fitting function in step (4.2) is carried out into the first order derivative multinomial that a derivation obtains fitting function, And solve X_iValue, by the X for solving_iValue substitutes into the maximum of T that fitting function can be obtained in fitting function_max, T_maxFor wind The peak of case EGT curve, as sintering end point temperature, X_iFor sintering end point position；

If the sintering end point temperature of matched curve is more than 300 DEG C, and the corresponding sintering of sintering end point temperature for obtaining is eventually Point position is located between penultimate bellows and third last bellows, then fitting function is reasonable, so that it is determined that fitting function With reasonability；

(4.4) the first order derivative multinomial of fitting function in step (4.3) is carried out into derivation, obtains the secondary of fitting function Derivative multinomial, and solve X_iValue；By the X for solving_iValue is substituted in fitting function and can obtain bellows high-temperature temperature value T_p, By T (X)=T_pSubstitute in fitting function and obtain X₁And X₂, wherein X₂>X₁；

(4.5) according to sintering mechanism, using the X solved in step (4.4)₁And X₂Value, using formula △ X=X₂- X₁, its difference △ X is obtained, high temperature hold time can be obtained using △ X, the computing formula of the high temperature hold time is as follows：

In formula：T represents high temperature hold time,Pallet average speed is represented, the pallet average speed is for Know value.

Further, in the step (5), sintering deposit yield prediction model is set up according to support vector regression algorithm.

Further, the support vector regression algorithm is set up sintering deposit yield prediction model and is comprised the following steps：

(5.1) sample data set of yield rate composition of Warm status parameter and sintering deposit is set as { (x_i, y_i), i=1,2 ... n},x_iFor |input paramete, y_iFor the output of corresponding target, the i.e. yield rate of sintering deposit；

(5.2) |input paramete is mapped to into higher dimensional space using Nonlinear Mapping, |input paramete is carried out in higher dimensional space Linear regression, and solve the optimization problem of linear regression problem：

In formula, ε is loss function parameter, and C is penalty factor, K (x_i·x_j) it is gaussian kernel function,α_iIt is weight coefficient；

(5.3) optimization problem in (5.2) is solved, obtaining sintering deposit yield prediction model is：

In formula：K(x_i, x)=exp (- | | x_i-x||²/2σ²), σ is kernel function width, and b is amount of bias, and the calculating of b values is public Formula is：

Further, it is described set up sintering deposit yield prediction model detailed process be：Randomly select and multigroup sinter The yield data and Warm status supplemental characteristic of ore deposit, by a part of finished product in multigroup yield data and Warm status supplemental characteristic Rate data and Warm status supplemental characteristic as training data, with a part of Warm status supplemental characteristic as input variable, with institute State a part of yield data to be computed repeatedly for output variable, set up sintering deposit yield prediction model；By multigroup finished product Remainder yield data and Warm status supplemental characteristic in rate data and Warm status supplemental characteristic as test data, with institute Remainder Warm status supplemental characteristic is stated for input variable, input sintering deposit yield prediction model, the sintering deposit yield rate The output variable of forecast model is the predicted value of yield rate, and the remainder yield data and the predicted value of yield rate are entered Row checking.

Compared with prior art, the invention has the advantages that：

(1) present invention is by the Analysis on Mechanism of sintering process, it is determined that the Warm status parameter of the yield rate of sintering deposit is affected, and Using bellows EGT data, the numerical value of Warm status parameter is obtained, be capable of achieving the yield rate for directly predicting sintering deposit, be burning Knot process real-time regulation technological parameter provides important evidence to improve sintering deposit yield rate and energy-saving consumption-reducing；

(2) present invention is based on Warm status parameter, and according to support vector regression algorithm sintering deposit yield prediction model is set up, Sintering deposit yield rate can directly be predicted, be effectively ensured forecast model it is accurate with it is reasonable；

(3) historical production data of the present invention based on sintering process, is predicted the emulation experiment of model, can be in reality Extensively apply in production process.

Description of the drawings

Fig. 1 is the sintering process process chart of strand exhaust sintering machine.

Fig. 2 is the schematic diagram of the sinter bed lamination of strand exhaust sintering machine.

Fig. 3 is the flow chart of one embodiment of the invention.

Fig. 4 is the bellows EGT matched curve figure of one embodiment of the invention.

Fig. 5 is difference △ X schematic diagrames on the bellows EGT curve of one embodiment of the invention.

Fig. 6 is the comparison diagram of the predicted value with the yield data of reality of the yield rate of one embodiment of the invention.

Fig. 7 is the error amount of the predicted value with the yield data of reality of the yield rate of one embodiment of the invention.

Specific embodiment

With reference to the accompanying drawings and examples the invention will be further described.

Using the invention provides a kind of sintering deposit yield prediction method based on bellows EGT, refer to Fig. 3, The present embodiment is comprised the following steps：

(1) the Warm status parameter of the yield rate for affecting sintering deposit is determined by analyzing the mechanism of the sintering process of sintering deposit, Sintering process includes solid phase reaction and liquid phase reactor, and liquid phase reactor can generate calcium ferrite liquid phase, and calcium ferrite liquid phase is sintering deposit The major influence factors of yield rate, and the generation of calcium ferrite liquid phase is mainly affected by sinter bed Warm status, while sintering The change of bed of material Warm status can be reflected by the change of bellows EGT curve, main for bellows EGT curve Characteristic parameter is bellows high-temperature temperature, high temperature hold time, sintering end point temperature and sintering end point position, therefore affects sintering deposit Yield rate Warm status parameter include bellows high-temperature temperature, high temperature hold time, sintering end point temperature and sintering end point position.

(2) historical production data to affecting the sintering parameter of sintering process is carried out at zero-phase filtering and sequential registration Reason, and 45min period of waves of selected great achievement ore deposit carries out sampling processing as the sampling period to historical production data, sets up history Creation data sample database；Sintering parameter includes bellows EGT, machine speed, little Cheng Kuang, returns mine and great achievement ore deposit；

In one embodiment, the detailed process for setting up historical production data sample database is as follows：Collect sintering machine one month Historical production data, gather each bellows EGT, machine speed, great achievement ore deposit, it is little into ore deposit and the history returned mine production number According to, due to there is uncertain factor, therefore historical production data in historical production data gatherer process in there is more hair Thorn, so as to need each box temperature to collecting in sintering process, machine speed, great achievement ore deposit, little into ore deposit and going through of returning mine History creation data carries out zero-phase filtering process；Then to each box temperature, machine speed, great achievement ore deposit, little into ore deposit and return mine Historical production data carry out sequential registration process, to guarantee that sintering process supplemental characteristic is consistent in sequential；Using big Into ore deposit period of waves 45min as the sampling period, to the historical production data through zero-phase filtering process and sequential registration Sampling processing is carried out, so as to set up historical production data sample database.

(3) yield rate of sintering deposit is calculated using historical production data sample database；The yield rate of sintering deposit is referred to After sintering process terminates, finished product sintering deposit accounts for the proportion of sinter cake, and the computing formula of the yield rate of sintering deposit is as follows：

In formula：ρ represents the yield rate (%) of sintering deposit, Q_DRepresent great achievement mineral products amount (Kg/h) of sintering, Q_XRepresent sintering It is little into mineral products amount (Kg/h), Q_FRepresent the quantity of return mines (Kg/h) of sintering；It is great achievement mineral products amount, little into mineral products amount and the data of quantity of return mines Obtain from historical production data sample database.

(4) using polynomial fitting method to the bellows EGT data march in historical production data sample database Line fitting obtains a fitting function, and by carrying out differential to fitting function extreme value is sought, and obtains the numerical value of Warm status parameter；

The numerical value for obtaining Warm status parameter specifically includes following steps：

(4.1) bellows EGT data sample is chosen from historical production data sample database, if bellows number is M, uses (X_i,T(X_i)) represent a sample data, i=1,2 ... M, X_iWith a distance from representing i-th bellows away from igniting, T (X_i) table Show with a distance from away from igniting to be X_iBellows exhaust gas temperature value；

(4.2) polynomial fitting method is adopted, to one group of sample data (X in bellows EGT data sample_i,T(X_i)) Fitting of a polynomial is carried out, fitting function is obtained for T (X_i)=a₈X_i ⁸+a₇X_i ⁷+a₆X_i ⁶+a₅X_i ⁵+a₄X_i ⁴+a₃X_i ³+a₂X_i ²+a₁X_i ¹+ a₀,

a₈、a₇、a₆、a₅、a₄、a₃、a₂、a₁And a₀Respectively it is fitted the coefficient for obtaining；

(4.3) fitting function in step (4.2) is carried out into the first order derivative multinomial that a derivation obtains fitting function, And solve X_iValue, by the X for solving_iValue substitutes into the maximum of T that fitting function can be obtained in fitting function_max, T_maxFor wind The peak of case EGT curve, as sintering end point temperature, X_iFor sintering end point position；

If the sintering end point temperature of matched curve is more than 300 DEG C, and the corresponding sintering of sintering end point temperature for obtaining is eventually Point position is located between penultimate bellows and third last bellows, then fitting function is reasonable, so that it is determined that fitting function With reasonability；

(4.4) the first order derivative multinomial of fitting function in step (4.3) is carried out into derivation, obtains the secondary of fitting function Derivative multinomial, and solve X_iValue；By the X for solving_iValue is substituted in fitting function and can obtain bellows high-temperature temperature value T_p, By T (X)=T_pSubstitute in fitting function and obtain X₁And X₂, wherein X₂>X₁；

(4.5) according to sintering mechanism, using the X solved in step (4.4)₁And X₂Value, using formula △ X=X₂- X₁, its difference △ X is obtained, high temperature hold time can be obtained using △ X, the computing formula of high temperature hold time is as follows：

In formula：T represents high temperature hold time,Pallet average speed is represented, pallet average speed is known Value.

Using above-mentioned steps, with reference to Fig. 2, a steel mill 360m²Sintering machine, has 24 bellows, between each bellows away from From difference, according to the actual conditions of this sintering machine, for scene be able to detect that 1^#、2^#、3^#、5^#、7^#、9^#、11^#、13^#、15^#、 17^#、18^#、19^#、20^#、21^#、22^#、23^#、24^#The exhaust gas temperature value of bellows, respectively correspond to pallet on away from igniting away from From be 1.5m, 4.5m, 7.5m, 14m, 22m, 30m, 38m, 46m, 54m, 62m, 66m, 70m, 74m, 78m, 82m, 85.5m, 88.5m；

One group of sample data (X is chosen in the bellows exhaust gas temperature value for detecting_i,T(X_i)), i=1,2 ... 17, its is concrete Be worth for (1.5,89.09154431), (4.5,58.20210398) ... (88.5,294.3742171), totally 17 groups, using multinomial Fitting process, it is 8 times to choose polynomial number of times, can solve fitting function T (X_i)=a₈X_i ⁸+a₇X_i ⁷+a₆X_i ⁶+a₅X_i ⁵+a₄X_i ⁴+ a₃X_i ³+a₂X_i ²+a₁X_i ¹+a₀Coefficient value a₈、a₇、a₆、a₅、a₄、a₃、a₂、a₁And a₀, the coefficient value of fitting function is through four houses five Numerical value after entering is as shown in the table, and bellows EGT matched curve figure is shown in Fig. 4.

The coefficient value of the fitting function after rounding up

a8

a7

a6

a5

a4

a3

a2

a1

a0

-2.69e-11

9.18e-9

-1.28e-6

9.19e-5

-0.004

0.071

-0.691

3.18

-181.6

The coefficient value of the fitting function for solving is updated to into fitting function

T(X_i)=a₈X_i ⁸+a₇X_i ⁷+a₆X_i ⁶+a₅X_i ⁵+a₄X_i ⁴+a₃X_i ³+a₂X_i ²+a₁X_i ¹+a₀In, fitting function is carried out once Derivation obtains the first order derivative multinomial of fitting function, solves below equation and X is obtained_iValue；

dT(X_i)/dX_i=d (a₈X_i ⁸+a₇X_i ⁷+a₆X_i ⁶+a₅X_i ⁵+a₄X_i ⁴+a₃X_i ³+a₂X_i ²+a₁X_i ¹+a₀)/dX_i=8a₈X_i ⁷+ 7a₇X_i ⁶+6a₆X_i ⁵+5a₅X_i ⁴+4a₄X_i ³+3a₃X_i ²+2a₂X_i ¹+a₁=0

By the X for solving_iValue is updated to fitting function

T(X_i)=a₈X_i ⁸+a₇X_i ⁷+a₆X_i ⁶+a₅X_i ⁵+a₄X_i ⁴+a₃X_i ³+a₂X_i ²+a₁X_i ¹+a₀In, that is, solve one group of sample number According in X_iT in ∈ [0m, 90m]_max, T_maxFor the peak of bellows EGT curve.Work as X by can be calculated_iWhen=83, T_maxMaximum is taken for 356.6533.Understand, sintering end point temperature is 356.6533 DEG C, sintering end point position is 83m；

Now sintering end point temperature is more than 300 DEG C and sintering end point position is located at penultimate bellows and third last Between bellows, illustrate that the fitting function is rational；In general, the maximum temperature of bellows EGT curve 300 DEG C with On.

The coefficient value of the fitting function for solving is substituted into into fitting function

T(X_i)=a₈X_i ⁸+a₇X_i ⁷+a₆X_i ⁶+a₅X_i ⁵+a₄X_i ⁴+a₃X_i ³+a₂X_i ²+a₁X_i ¹+a₀In, once leading to fitting function Number multinomial carries out derivation, obtains the second derivative multinomial of fitting function, and solves the X of below equation_iValue；

56a₈X_i ⁶+42a₇X_i ⁵+30a₆X_i ⁴+20a₅X_i ³+12a₄X_i ²+6a₃X_i+2a₂=0

By the X for solving_iValue substitutes into fitting function

T(X_i)=a₈X_i ⁸+a₇X_i ⁷+a₆X_i ⁶+a₅X_i ⁵+a₄X_i ⁴+a₃X_i ³+a₂X_i ²+a₁X_i ¹+a₀In, bellows high-temperature temperature value can be obtained T_p=259.4902, then by T (X)=T_p=259.4902 substitute into fitting function

T(X_i)=a₈X_i ⁸+a₇X_i ⁷+a₆X_i ⁶+a₅X_i ⁵+a₄X_i ⁴+a₃X_i ³+a₂X_i ²+a₁X_i ¹+a₀In try to achieve X₁And X₂, X can be obtained₁= 70.800, X₂=89.8207, using formula △ X=X₂-X₁, obtain its difference △ X=89.8207-70.800=19.0207； Difference △ X schematic diagrames are shown in Fig. 5 on bellows EGT curve；

Because high temperature hold time is equal to difference △ X divided by pallet average speed, so being solved according to above-mentioned △ X and known pallet average speed, can obtain high temperature hold time t, and it is expressed as：

(5) numerical value of the Warm status parameter obtained using step (4) is used as input variable, with the sintering that step (3) is obtained The yield rate of ore deposit is computed repeatedly as output variable and verified, according to support vector regression algorithm sintering deposit yield rate is set up Forecast model；

Concretely comprising the following steps for sintering deposit yield prediction model is set up according to support vector regression algorithm：

(5.1) sample data set of yield rate composition of Warm status parameter and sintering deposit is set as { (x_i,y_i), i=1,2 ... n},x_iFor |input paramete, y_iFor the output of corresponding target, the i.e. yield rate of sintering deposit；

(5.2) |input paramete is mapped to into higher dimensional space using Nonlinear Mapping, |input paramete is carried out in higher dimensional space Linear regression, and solve the optimization problem of linear regression problem：

In formula, ε is loss function parameter, and C is penalty factor, K (x_i·x_j) it is gaussian kernel function,α_iIt is weight coefficient；

(5.3) optimization problem in (5.2) is solved, obtaining sintering deposit yield prediction model is：

In formula：K(x_i, x)=exp (- | | x_i-x||²/2σ²), σ is kernel function width, and b is amount of bias, and the calculating of b values is public Formula is：

(6) the bellows EGT data of ore deposit to be predicted are carried out curve fitting and by differential using polynomial fitting method Extremum method is asked to obtain the numerical value of the Warm status parameter of ore deposit to be predicted；

(7) sintering deposit yield prediction model is input into using the numerical value of the Warm status parameter of ore deposit to be predicted as input variable, The output variable of sintering deposit yield prediction model is the yield rate of ore deposit to be predicted.

In one embodiment, 200 groups of yield data of sintering deposit and Warm status supplemental characteristics are randomly selected, with 170 groups Yield data and Warm status supplemental characteristic as training data, with Warm status supplemental characteristic as input variable, with yield rate number According to being computed repeatedly for output variable, sintering deposit yield prediction model is set up；

Using 30 groups of yield datas and Warm status supplemental characteristic as test data, become by input of Warm status supplemental characteristic Amount, is input into sintering deposit yield prediction model, and the output variable of sintering deposit yield prediction model is the predicted value of yield rate, will Yield data is verified that predicted value is divided with the comparison diagram and error amount of the yield data of reality with the predicted value of yield rate Do not see Fig. 6 and Fig. 7, as shown in Figure 7, the relative error for predicting the outcome of the yield rate of sintering deposit [- 0.06%, 0.08%] it It is interior, therefore, the sintering deposit yield prediction model of foundation has feasibility.

This method can realize directly predicting the yield rate of sintering deposit that predictablity rate is high, is sintering process real-time regulation Technological parameter provides important evidence to improve sintering deposit yield rate and energy-saving consumption-reducing.

In the case where not conflicting, the feature in embodiment herein-above set forth and embodiment can be combined with each other.

The foregoing is only presently preferred embodiments of the present invention, not to limit the present invention, all spirit in the present invention and Within principle, any modification, equivalent substitution and improvements made etc. should be included within the scope of the present invention.

Claims (9)

1. a kind of sintering deposit yield prediction method based on bellows EGT, it is characterised in that：Comprise the following steps：

(1) the Warm status parameter of the yield rate for affecting sintering deposit is determined according to the sintering process of sintering deposit；

(2) historical production data to affecting the sintering parameter of sintering process carries out zero-phase filtering and sequential registration process, and Sampling processing is carried out to the historical production data, historical production data sample database is set up；

(3) yield rate of sintering deposit is calculated using the historical production data sample database；

(4) using polynomial fitting method to the bellows EGT data march in the historical production data sample database Line fitting obtains a fitting function, and by carrying out differential to the fitting function extreme value is sought, and obtains the numerical value of Warm status parameter；

(5) numerical value of the Warm status parameter obtained using step (4) is used as input variable, the sintering deposit obtained with step (3) Yield rate is computed repeatedly as output variable and verified, sets up sintering deposit yield prediction model；

(6) using polynomial fitting method the bellows EGT data of ore deposit to be predicted are carried out curve fitting and asks pole by differential Value method obtains the numerical value of the Warm status parameter of ore deposit to be predicted；

(7) numerical value of the Warm status parameter of the ore deposit to be predicted for obtaining step (6) is input into sintering deposit yield rate as input variable Forecast model, the output variable of the sintering deposit yield prediction model is the yield rate of ore deposit to be predicted.

2. the sintering deposit yield prediction method of bellows EGT is based on as claimed in claim 1, it is characterised in that described In step (1), Warm status parameter is obtained by analyzing the mechanism of sintering process, and analysis determines the heat of the yield rate for affecting sintering deposit State parameter includes bellows high-temperature temperature, high temperature hold time, sintering end point temperature and sintering end point position.

3. the sintering deposit yield prediction method of bellows EGT is based on as claimed in claim 1, it is characterised in that described Sintering parameter includes bellows EGT, machine speed, little Cheng Kuang, returns mine and great achievement ore deposit.

4. the sintering deposit yield prediction method of bellows EGT is based on as claimed in claim 1, it is characterised in that described In step (2), the selected sampling period carries out sampling processing to historical production data, and the sampling period is all for the fluctuation of great achievement ore deposit Phase, the period of waves of the great achievement ore deposit is 45min.

5. the sintering deposit yield prediction method of bellows EGT is based on as claimed in claim 1, it is characterised in that sintering After the yield rate of ore deposit refers to that sintering process terminates, finished product sintering deposit accounts for the proportion of sinter cake, the meter of the yield rate of the sintering deposit Calculate formula as follows：

$\mathrm{\ρ}=\frac{Q_D+Q_X}{Q_D+Q_X+Q_F};$

In formula：ρ represents the yield rate (%) of sintering deposit, Q_DRepresent great achievement mineral products amount (Kg/h) of sintering, Q_XRepresent sintering it is little into Mineral products amount (Kg/h), Q_FRepresent the quantity of return mines (Kg/h) of sintering；It is the great achievement mineral products amount, little into mineral products amount and the data of quantity of return mines Obtain from historical production data sample database.

6. a kind of sintering deposit yield prediction method based on bellows EGT as claimed in claim 2, it is characterised in that The numerical value for obtaining Warm status parameter is comprised the following steps：

(4.1) bellows EGT data sample is chosen from the historical production data sample database, if bellows number is M, uses (X_i,T(X_i)) represent a sample data, i=1,2 ... M, X_iWith a distance from representing i-th bellows away from igniting, T (X_i) table Show with a distance from away from igniting to be X_iBellows exhaust gas temperature value；

(4.2) polynomial fitting method is adopted, to one group of sample data (X in bellows EGT data sample_i,T(X_i)) carry out Fitting of a polynomial, obtains fitting function for T (X_i)=a₈X_i ⁸+a₇X_i ⁷+a₆X_i ⁶+a₅X_i ⁵+a₄X_i ⁴+a₃X_i ³+a₂X_i ²+a₁X_i ¹+a₀,

a₈、a₇、a₆、a₅、a₄、a₃、a₂、a₁And a₀Respectively it is fitted the coefficient for obtaining；

(4.3) fitting function in step (4.2) is carried out into a derivation and obtains the first order derivative multinomial of fitting function, and asked Solve X_iValue, by the X for solving_iValue substitutes into the maximum of T that fitting function can be obtained in fitting function_max, T_maxIt is useless for bellows The peak of gas temperature curve, as sintering end point temperature, X_iFor sintering end point position；

If the sintering end point temperature of matched curve is more than 300 DEG C, and the corresponding sintering end point position of sintering end point temperature for obtaining Setting between penultimate bellows and third last bellows, then fitting function is reasonable, so that it is determined that fitting function has Reasonability；

(4.4) the first order derivative multinomial of fitting function in step (4.3) is carried out into derivation, obtains the second derivative of fitting function Multinomial, and solve X_iValue；By the X for solving_iValue is substituted in fitting function and can obtain bellows high-temperature temperature value T_p, by T (X)=T_pSubstitute in fitting function and obtain X₁And X₂, wherein X₂>X₁；

(4.5) according to sintering mechanism, using the X solved in step (4.4)₁And X₂Value, using formula △ X=X₂-X₁, obtain Its difference △ X, using △ X high temperature hold time can be obtained, and the computing formula of the high temperature hold time is as follows：

$t=\frac{\mathrm{\Δ}X}{\stackrel{\‾}{v}};$

In formula：T represents high temperature hold time,Pallet average speed is represented, the pallet average speed is known Value.

7. the sintering deposit yield prediction method of bellows EGT is based on as claimed in claim 1, it is characterised in that described In step (5), sintering deposit yield prediction model is set up according to support vector regression algorithm.

8. the sintering deposit yield prediction method of bellows EGT is based on as claimed in claim 7, it is characterised in that described Support vector regression algorithm is set up sintering deposit yield prediction model and is comprised the following steps：

(5.1) sample data set of yield rate composition of Warm status parameter and sintering deposit is set as { (x_i,y_i), i=1,2 ... n }, x_i For |input paramete, y_iFor the output of corresponding target, the i.e. yield rate of sintering deposit；

(5.2) |input paramete is mapped to into higher dimensional space using Nonlinear Mapping, |input paramete is carried out linearly in higher dimensional space Return, and solve the optimization problem of linear regression problem：

$\left\{\begin{array}{c}\min \{-\frac{1}{2}\underset{i,j=1}{\overset{n}{\Σ}}({\mathrm{\α}}_i^{*}-{\mathrm{\α}}_i)({\mathrm{\α}}_j^{*}-{\mathrm{\α}}_j)K(x_i\·x_j)+\mathrm{\ϵ}\underset{i=1}{\overset{n}{\Σ}}({\mathrm{\α}}_i^{*}+{\mathrm{\α}}_i)-\underset{i=1}{\overset{n}{\Σ}}{y}_i({\mathrm{\α}}_i^{*}-{\mathrm{\α}}_i)\}\\ \begin{array}{cccc}s.t.& \underset{i=1}{\overset{n}{\Σ}}({\mathrm{\α}}_i^{*}-{\mathrm{\α}}_i)=0,& 0\≤{\mathrm{\α}}_i^{*},{\mathrm{\α}}_i\≤C,& i=1,2,\mathrm{...},n\end{array}\end{array}\right.$

In formula, ε is loss function parameter, and C is penalty factor, K (x_i·x_j) it is gaussian kernel function,α_iIt is weight coefficient；

(5.3) optimization problem in (5.2) is solved, obtaining sintering deposit yield prediction model is：

$y=\underset{i=1}{\overset{n}{\Σ}}({{\mathrm{\α}}_i}^{*}-{\mathrm{\α}}_i)K(x_i,x)+b$

In formula：K(x_i, x)=exp (- | | x_i-x||²/2σ²), σ is kernel function width, and b is amount of bias, and the computing formula of b values is：

$b=\left\{\begin{array}{cc}{y}_i-\underset{j=1}{\overset{n}{\Σ}}({{\mathrm{\α}}_j}^{*}-{\mathrm{\α}}_j)K(x_i,x_j)+\mathrm{\ϵ},& {\mathrm{\α}}_j\∈(0,C)\\ y_i-\underset{j=1}{\overset{n}{\Σ}}({{\mathrm{\α}}_j}^{*}-{\mathrm{\α}}_j)K(x_i,x_j)-\mathrm{\ϵ},& {{\mathrm{\α}}_j}^{*}\∈(0,C)\end{array}\right.$

9. the sintering deposit yield prediction method of bellows EGT is based on as claimed in claim 7, it is characterised in that described The detailed process for setting up sintering deposit yield prediction model is：Randomly select the yield data and Warm status of multigroup sintering deposit Supplemental characteristic, by a part of yield data and Warm status supplemental characteristic in multigroup yield data and Warm status supplemental characteristic As training data, with a part of Warm status supplemental characteristic as input variable, with a part of yield data as defeated Go out variable to be computed repeatedly, set up sintering deposit yield prediction model；By multigroup yield data and Warm status supplemental characteristic In remainder yield data and Warm status supplemental characteristic as test data, with the remainder Warm status parameter number According to for input variable, sintering deposit yield prediction model is input into, the output variable of the sintering deposit yield prediction model is into The predicted value of product rate, the remainder yield data and the predicted value of yield rate are verified.