CN108171381A - A kind of blast furnace CO utilization rates chaos weighing first order local prediction method and system - Google Patents
A kind of blast furnace CO utilization rates chaos weighing first order local prediction method and system Download PDFInfo
- Publication number
- CN108171381A CN108171381A CN201711483980.5A CN201711483980A CN108171381A CN 108171381 A CN108171381 A CN 108171381A CN 201711483980 A CN201711483980 A CN 201711483980A CN 108171381 A CN108171381 A CN 108171381A
- Authority
- CN
- China
- Prior art keywords
- blast furnace
- utilization rate
- carbon monoxide
- phase space
- time series
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
Abstract
The invention discloses a kind of blast furnace utilization rate of carbon monoxide chaos weighing first order local prediction method and systems, phase space insertion time and embedded dimension are determined by using auto-relativity function method and correlation integral method, and phase space reconfiguration is carried out to blast furnace utilization rate of carbon monoxide time series using delay coordinate method;Denoising is carried out to blast furnace utilization rate of carbon monoxide time series using Wavelet noise-eliminating method;Blast furnace utilization rate of carbon monoxide chaos weighing first order local increment is established, and calculates the coefficient in first-order linear model of fit;According to fixed optimal coefficient, blast furnace utilization rate of carbon monoxide predicted value is obtained.The present invention provides effective prediction for blast furnace utilization rate of carbon monoxide, solves the optimization of blast furnace energy consumption index, reduces the purpose of energy consumption.
Description
Technical field
The invention belongs to blast furnace utilization rate of carbon monoxide ultra-short term electric powder predictions, are related to a kind of blast furnace CO utilization rates
Chaos weighing first order local prediction method and system.
Background technology
Steel and iron industry is the pillar industries of the national economy, meanwhile, also belong to high energy consumption and the industry of maximum discharge.With resource
Economizing type, the proposition of environmentally friendly social goal and development accelerate, and steel and iron industry bears the weight of more energy-saving and emission-reduction
Appoint, so as to fulfill low consumption environmental protection.
As component part main and most important in smelting iron and steel, blast furnace is not only the key equipment smelted, simultaneously
It is also the energy, resource consumption and the rich and influential family of pollution in iron and steel enterprise.In blast furnace ironmaking process, blast furnace utilization rate of carbon monoxide is straight
The utilization of the energy consumption for influencing ton iron and high capacity of furnace is connect, the quality of blast furnace using energy source can be assessed well;Meanwhile blast furnace one
Aoxidizing carbon utilisation rate can be with the current blast furnace operating level of real-time characterization and the situation of operation.Therefore, to blast furnace energy consumption index
Utilization rate of carbon monoxide accurately predict and control, and real-time guide is provided for blast furnace execute-in-place, so as to ensure blast furnace
Steady operation, have important practical significance.
However, seal and complexity due to blast furnace working environment, it is generally difficult to blast furnace one be described by foundation and aoxidized
The state equation or mechanism model of carbon utilisation rate development and change analyzes blast furnace utilization rate of carbon monoxide, and then grasps its development and become
The rule of change accurately predicts it.In addition, at operation scene, the prediction of usual blast furnace utilization rate of carbon monoxide is main
It is or limited due to artificial experience by the experience of site operation personnel so as to predict the variation hair of utilization rate of carbon monoxide
Exhibition trend has certain subjectivity and limitation.Due to lacking the other methods such as data-driven to blast furnace utilization rate of carbon monoxide
Forecast analysis, cause Field Force that cannot carry out commenting in real time well to blast furnace utilization rate of carbon monoxide development tendency
Estimate, so as to miss the adjustment of execute-in-place, the synthesis improvement of utilization rate of carbon monoxide in actual production is caused to be restricted.Institute
With development & construction blast furnace utilization rate of carbon monoxide forecasting system, blast furnace utilization rate of carbon monoxide is precisely predicted, for scene
The operation for optimizing blast furnace utilization rate of carbon monoxide provides guidance, realizes the purpose of blast furnace consumption reduction, it appears particularly important.
Invention content
For existing blast furnace utilization rate of carbon monoxide prediction model and the deficiency of prediction accuracy, a kind of one oxygen of blast furnace is provided
Change carbon utilisation rate chaos weighing first order local prediction method and system, to solve the optimization of blast furnace energy consumption index, reduce the mesh of energy consumption
's.
To achieve these goals, the present invention provides a kind of blast furnace CO utilization rates chaos weighing first order local prediction sides
Method includes the following steps:
The time series η (co) of S1, the original blast furnace utilization rate of carbon monoxide of inputi, i=1,2 ..., N, wherein N are blast furnace
The length of utilization rate of carbon monoxide sample time-series;
S2, height is determined using auto-relativity function method to the time series of original blast furnace utilization rate of carbon monoxide inputted in S1
Stove utilization rate of carbon monoxide time series phase space is embedded in time τ;
S3, blast furnace is determined using correlation integral method to the time series of original blast furnace utilization rate of carbon monoxide inputted in S1
Utilization rate of carbon monoxide time series phase space Embedded dimensions m;
S4, using in step S2 and S3 determine insertion time two parameters of τ and Embedded dimensions m, using delay coordinate method
Phase space reconfiguration is carried out to blast furnace utilization rate of carbon monoxide time series;
S5, denoising is carried out to the time series phase space reconstruction obtained in step S4 using Wavelet noise-eliminating method;
S6, according to the blast furnace utilization rate of carbon monoxide time series phase space after denoising in step S5, it is mutually empty to choose reconstruct
Between the last one phase point as prediction central point ZM, choose and prediction central point ZMDistance is siPoint be adjacent phase space point ZMi,
Determine adjacent phase space point ZMiWeights Qi;
S7, the adjacent phase space point Z obtained according to step S6MiWeights Qi, establish blast furnace utilization rate of carbon monoxide chaos
Local increment is weighted, and calculates the coefficient matrix in first-order linear model of fit;
S8, according in step S7 determine optimal coefficient matrix, obtain blast furnace utilization rate of carbon monoxide predicted value.
It is based in blast furnace CO utilization rate chaos weighing first order local prediction methods in one kind of the present invention, adopting in step S2
Determine that the embedded time is realized by following steps with auto-relativity function method:
S21, the auto-correlation function for establishing blast furnace utilization rate of carbon monoxide time series
S22, the auto-correlation function expression formula according to S21, work as Cττ values when value is intended to 0 can be identified as blast furnace one
Aoxidize the insertion time τ of carbon utilisation rate phase space reconstruction.
It is based in blast furnace CO utilization rate chaos weighing first order local prediction methods in one kind of the present invention, the pass in step S3
Connection integration method determines that embedded dimension is realized by following steps:
S31, the number M for obtaining blast furnace utilization rate of carbon monoxide phase space reconstruction midpoint;
S32, the correlation integral functional equation for establishing blast furnace utilization rate of carbon monoxide sequential
Wherein, dijRepresent the Euclidean distance between arbitrary two vectors point, r in blast furnace utilization rate of carbon monoxide phase space reconstruction
For distance optional in phase space,For Heaviside functions;
Correlation integral functional equation in S33, the numerical value and S32 of the insertion time τ that are determined according to step S2, obtains blast furnace
The calculation expression of utilization rate of carbon monoxide correlation dimension D:
S34, it can be calculated by the correlation dimension in step S33, the lnC of blast furnace sample space(co)m(r) with the slope of lnr
Change rate be less than preset threshold value k when, m be smallest embedding dimension degree, so as to obtain blast furnace utilization rate of carbon monoxide time series
Phase space reconstruction Embedded dimensions m.
It is based in blast furnace CO utilization rate chaos weighing first order local prediction methods in one kind of the present invention, adopting in step S4
Phase space reconfiguration is carried out to blast furnace utilization rate of carbon monoxide time series with delay coordinate method, is obtained as follows:
S41:Input utilization rate of carbon monoxide phase space reconstruction insertion time τ and Embedded dimensions m;
S42:Phase space reconfiguration is carried out using delay coordinate method, the phase space after reconstruct is specially:
Wherein, N is the number of utilization rate of carbon monoxide original sample time series, and M=N- (m-1) τ is carbon monoxide profit
With the vectorial number of rate time series embedding phase space.
It is based in blast furnace CO utilization rate chaos weighing first order local prediction methods in one kind of the present invention, in the step S6
Weights QiSpecifically calculating process is:
It is calculated by equation below and determines adjacent phase space point ZMiWeights Qi;
Wherein, h is constant, usually takes the number that 1, p is consecutive points phase space point, siFor adjacent phase space point ZMiTo prediction
Central point ZMDistance, sminFor siIn minimum value.
It is based in blast furnace CO utilization rate chaos weighing first order local prediction methods in one kind of the present invention, the step S7 tools
Body step is:
S71:According to the weights Q of adjacent space point obtained in step S6i, establish blast furnace utilization rate of carbon monoxide chaos one
Rank weights local increment, establishes adjacent phase space point ZMiWith kth step evolutionary phase spatial point ZMi+kBetween model;
ZMi+k=cke+gkZMi, i=1,2 ..., N
S72:Using following weighted least-squares method linear equation, weighted least-squares fitting linear function T is established
With kth step evolutionary phase spatial point ZMi+kBetween model;
Wherein,It is vector ZMiJ-th of element;
S73:Local derviation is asked to calculate simultaneously on the weighted least-squares method linear equation both sides obtained in step S72, really
Determine optimum linearity fitting coefficient matrix ckAnd gkValue.
It is based in blast furnace CO utilization rate chaos weighing first order local prediction methods in one kind of the present invention, the height in step S7
Stove utilization rate of carbon monoxide predicted value can be iterated to calculate by following formula and be obtained:
ZM+k=cke+gkZM
Wherein, e=(1,1 ..., 1) T is p dimensional vectors, ZM+kBlast furnace utilization rate of carbon monoxide predicted value is walked for kth.
It is based in blast furnace CO utilization rate chaos weighing first order local prediction methods, further includes in one kind of the present invention:Pass through
The blast furnace utilization rate of carbon monoxide prediction that step S8 is obtained^Value calculates the precision of prediction of model as follows;
Obtain blast furnace utilization rate of carbon monoxide predicted value z (m), blast furnace utilization rate of carbon monoxide actual value z (m), prediction step
Number k;
According to formulaWithIt calculates
Opposite error (the P based on chaos weighing first order partial modelerr) and root-mean-square error (RMSE);
Relative error (the P calculated according to two above formulaerr) and root-mean-square error (RMSE) weigh based on chaos
The precision of prediction of weighing first order partial model.
Preferably, it, based on blast furnace CO utilization rate chaos weighing first order local prediction systems, is wrapped the invention also includes one kind
It includes with lower module:
Just enter initialization data information module, for inputting the time series η of original blast furnace utilization rate of carbon monoxide
(co)i, i=1,2 ..., N, wherein N are the length of blast furnace utilization rate of carbon monoxide sample time-series;
Embedded time module is calculated, for the original blast furnace utilization rate of carbon monoxide of initialization data information module input
Time series determines that blast furnace utilization rate of carbon monoxide time series phase space reconstruction is embedded in time τ using auto-relativity function method;
Embedded dimensions module is calculated, association is used for the time series of the original blast furnace utilization rate of carbon monoxide to input
Integration method determines blast furnace utilization rate of carbon monoxide time series phase space reconstruction Embedded dimensions m;
Phase space reconfiguration module, it is mutually empty for being carried out using delay coordinate method to blast furnace utilization rate of carbon monoxide time series
Between reconstruct;
Wavelet Denoising Method module, for using Wavelet noise-eliminating method to the utilization rate of carbon monoxide time sequence after phase space reconfiguration
Row carry out denoising;
Weight computing module, it is mutually empty according to the blast furnace utilization rate of carbon monoxide time series after Wavelet Denoising Method module denoising
Between, determine prediction central point ZM, find adjacent phase space point Z corresponding with the central pointMi, determine adjacent phase space point ZMiPower
Value Qi;
Model coefficient matrix acquisition module, for the adjacent phase space point Z obtained according to previous stepMiWeights Qi, establish
Blast furnace utilization rate of carbon monoxide chaos weighing first order local increment, and calculate the coefficient square in first-order linear model of fit
Battle array;
Predicted value acquisition module according to determining optimal coefficient matrix, obtains blast furnace utilization rate of carbon monoxide predicted value.
Compared with prior art, the advantageous effect of present invention is that:
1. starting with from the angle of time series, blast furnace utilization rate of carbon monoxide is carried out using chaos phase space technology pre-
It surveys, avoids the influence of other various external factors, ensure that the precision of prediction.
2. using wavelet technique to come from scene blast furnace utilization rate of carbon monoxide sample data denoising, ensure
The reliability of experimental data, establishes accurate prediction model for lower step and lays a good foundation.
3. the blast furnace utilization rate of carbon monoxide chaos weighing first order local increment established is simple, calculation amount is small, takes
It is short, ensure that prediction result is accurate.
4. the chaos weighing first order local prediction of blast furnace utilization rate of carbon monoxide, the practical operation for scene provides effectively
It guides, realizes the optimization of blast furnace energy consumption index, the purpose for reducing energy consumption.
Description of the drawings
Below in conjunction with accompanying drawings and embodiments, the invention will be further described, in attached drawing:
Fig. 1 is general flow chart of the present invention;
Fig. 2 is 1100m of the present invention3Blast furnace utilization rate of carbon monoxide sample time-series figure;
Fig. 3 is 3200m of the present invention3Blast furnace utilization rate of carbon monoxide sample time-series figure;
Fig. 4 is 1100m of the present invention3Blast furnace utilization rate of carbon monoxide time series autocorrelation function graph;
Fig. 5 is 3200m of the present invention3Blast furnace utilization rate of carbon monoxide time series autocorrelation function graph;
Fig. 6 is 1100m of the present invention3Blast furnace utilization rate of carbon monoxide time series correlation integral figure;
Fig. 7 is 3200m of the present invention3Blast furnace utilization rate of carbon monoxide time series correlation integral figure;
Fig. 8 is 1100m of the present invention3Blast furnace utilization rate of carbon monoxide Wavelet Denoising Method figure;
Fig. 9 is 3200m of the present invention3Blast furnace utilization rate of carbon monoxide Wavelet Denoising Method figure;
Figure 10 is 1100m of the present invention3Prognostic chart before blast furnace utilization rate of carbon monoxide denoising;
Figure 11 is 1100m of the present invention3Prognostic chart after blast furnace utilization rate of carbon monoxide denoising;
Figure 12 is 3200m of the present invention3Prognostic chart before blast furnace utilization rate of carbon monoxide denoising;
Figure 13 is 3200m of the present invention3Prognostic chart after blast furnace utilization rate of carbon monoxide denoising.
Specific embodiment
In order to make the purpose , technical scheme and advantage of the present invention be clearer, below in conjunction with attached drawing and example, to this
Invention is further elaborated.
With reference to the blast furnace utilization rate of carbon monoxide sample time-series data of actual samples, Fig. 1 is shown in using flow chart of the present invention, this
Implement the specific steps are:
In the present embodiment, two groups are had chosen under different working conditions, and two seat tools from prominent domestic steel plant have
Representative middle high type blast furnace A factory 1100m3No. 7 blast furnaces and B factories 3200m3No. 2 blast furnaces, with the blast furnace 2012 2 of A factories 7
Utilization rate of carbon monoxide ((η the co)) time series and the carbon monoxide in the blast furnace of B factories 2 in May, 2012 of month acquisition utilize
For rate ((η co)) time series as sample space, capacity is 5000 data, and the sampling period is 3 minutes, and sample time-series are shown in figure
2 and Fig. 3.
The insertion time of blast furnace utilization rate of carbon monoxide phase space reconstruction is determined using auto-relativity function method:
For the time series η (co) of the utilization rate of carbon monoxide of acquisitioni, i=1,2 ..., N, N is blast furnace carbon monoxide
The length of utilization rate sample time-series.
Then the auto-correlation function of blast furnace utilization rate of carbon monoxide time series is defined as:
The auto-correlation function simulation figure of two seat height stove utilization rate of carbon monoxide is respectively Fig. 4 and Fig. 5.It can be with from Fig. 4
Find out, when τ=10, the auto-correlation function coefficient C of blast furnace utilization rate of carbon monoxideτClosest to 0 point;Similarly in Figure 5 can also
Obtain same result.So as to us it is found that this two typical blast furnace utilization rate of carbon monoxide retiming phase spaces when
The stagnant time is 10.
The Embedded dimensions of blast furnace utilization rate of carbon monoxide phase space reconstruction are determined using correlation dimension method:
For the sample time-series of original blast furnace utilization rate of carbon monoxide
(ηco)i={ (η co)1,(ηco)2,(ηco)3,…,(ηco)N, i=1,2,3 ... N
The correlation integral of two seat height stove utilization rate of carbon monoxide sequential is:
Numbers of the M for blast furnace utilization rate of carbon monoxide phase space reconstruction midpoint, d in above formulaijRepresent blast furnace carbon monoxide profit
With the Euclidean distance between two vectors point arbitrary in rate phase space reconstruction, r is distance optional in phase space.θ () is
Heaviside functions.Wherein
It, can be to utilization rate of carbon monoxide phase space reconstruction using correlation dimension method according to the numerical value of insertion time τ asked
Correlation integral calculated.
When M is sufficiently large, during and r → 0, correlation integral C(co)m(r) there are following relational expressions with r:
Wherein D is the correlation dimension of blast furnace utilization rate of carbon monoxide phase space reconstruction, the numerical value of appropriate selection r so that D
The self-similar structure of strange attractor, therefore the calculation expression of correlation dimension as available from the above equation can be portrayed:
The correlation dimension analogous diagram of two seat height stove utilization rate of carbon monoxide time serieses is Fig. 6 and Fig. 7, can be obtained by two figures,
When m is 7 and 9 respectively, 1100m3And 3200m3The lnC of two seat height stove sample spaces(co)m(r) it has been protected substantially with the slope of lnr
It holds constant.Therefore, m=7 can be considered 1100m3The smallest embedding dimension number of the phase space reconstruction of blast furnace;M=9 can be considered 3200m3It is high
The smallest embedding dimension number of the phase space reconstruction of stove.
According to acquired embedded time and the numerical value of Embedded dimensions, blast furnace carbon monoxide is utilized using delay coordinate method
Rate time series carries out phase space reconfiguration, and phase space expression is:
In this example, to the live blast furnace utilization rate of carbon monoxide sample data of two typicalness blast furnaces:That is A factories 7
The one of utilization rate of carbon monoxide ((η co)) time series and No. 2 blast furnaces of B factories in May, 2012 of blast furnace 2 months in 2012 acquisition
Carbon utilisation rate ((η co)) time series is aoxidized, using principle of wavelet analysis, Wavelet Denoising Method is carried out to it, is emulated by Matlab
Afterwards, figure is as shown in FIG. 8 and 9.
From two it can be seen from the figure thats, on the basis of original sample clock signal, after carrying out Wavelet Denoising Method, original temporal is believed
Burr and spike in number have obtained effective removal, while original feature has obtained good guarantor in sample time-series signal
It stays, denoising effect is more satisfactory
Phase space after being reconstructed based on blast furnace utilization rate of carbon monoxide chooses the last one phase point of phase space reconstruction as pre-
Measured center point ZM, choose prediction central point ZMConsecutive points ZMiI=1,2 ..., N have similar changing rule to central point.
And arrive ZMDistance be siIf sminFor siIn minimum value, calculate point ZMiWeights be
Based on weight computing as a result, establishing blast furnace utilization rate of carbon monoxide chaos weighing first order Local Linear Model, calculate
Optimum linearity fitting coefficient.
Consecutive points ZMiIt is { Y with the phase point set after its k step that developMi+k, therefore first order local area linear fit is:
ZMi+k=cke+gkZMi, i=1,2 ..., N
It can be obtained by weighted least-squares method:
WhereinIt is vector ZMiJ-th of element.Obvious above formula is ckAnd gkBinary function, above formula both sides are asked simultaneously
Local derviation can obtain
Above formula is subjected to abbreviation, can be obtained:
Above formula is converted into matrix is:
Wherein
The best linear fit coefficient obtained according to previous step, the blast furnace utilization rate of carbon monoxide predicted value can pass through
Following formula, which calculates, to be obtained:
ZM+k=cke+gkZM(e=(1,1 ..., 1)TFor p dimensional vectors, 1) k values is
Wherein, ZM+kBlast furnace utilization rate of carbon monoxide time series phase space point is walked for k, when k values are 1, the blast furnace
Utilization rate of carbon monoxide one-step prediction value is ZM+1+(m-1)τ。
In addition, relative error (P is also chosen in this exampleerr) and root-mean-square error (RMSE) come evaluation model prediction essence
Degree.
Relative error (Perr) definition is as follows:
The definition of root-mean-square error (RMSE) is as follows:
Wherein,It is utilization rate of carbon monoxide predicted value, z (m) is then its actual value, and k is that it predicts step number.
Blast furnace utilization rate of carbon monoxide chaotic prediction mould is built using chaos weighing first order local area Forecasting Methodology in this example
Type predicts rear 200 groups of data of two seat height stoves, after using preceding 5000 data predict after phase space reconfiguration
240 data.Prediction result such as Fig. 9~Figure 13.Wherein, Fig. 9~Figure 10 is 1100m3The prediction result of blast furnace (before denoising and is gone
After making an uproar), and Figure 12~Figure 13 is 3200m3The prediction result of blast furnace (before denoising and after denoising);Prediction errors table is table 1.
(chaos weighing first order local area predicts mould to the prediction errors table of 1 two sample blast furnace utilization rate of carbon monoxide sequential of table
Type)
It can be seen from simulation figure and table on the one hand:It can be seen from the figure of Figure 11 and Figure 13 (after denoising)
When doing chaos weighing first order local prediction to the practical sequential of blast furnace utilization rate of carbon monoxide, predicted value can follow reality well
Precision of prediction is very high in the variation of value, especially Figure 13, so that the curve of actual value can be completely covered in predicted value in figure.
Similar conclusion also can be by observing 1100m3The prediction figure of blast furnace utilization rate of carbon monoxide obtains.On the other hand:It is in addition, logical
The prediction curve compared before denoising and after denoising is crossed, it should be apparent that the prediction error before denoising is than the prediction after denoising
Error is big, that is, the precision of prediction higher after denoising, so as to illustrate the necessity of denoising.
In addition, the conclusion that analysis prediction figure obtains can also equally be obtained by the date comprision in errors table,
Specific error value is shown in Table 1.2nd, 3 line number values of table represent 1100m respectively3The chaos weighing first order local area of data before and after denoising
The relative error P of prediction resulterrWith root-mean-square error RMSE.And the 4th, 5 line number values then represent 3200m respectively3Number before and after denoising
According to prediction result relative error PerrWith root-mean-square error RMSE.It can easily be seen that by comparison, 1100m3Blast furnace is gone
Make an uproar front and rear relative error PerrRespectively 0.0060 and 0.0045, root-mean-square error RMSE are respectively 0.0031 and 0.0028,
25% and 10% are reduced respectively after two kinds of error value denoisings.And 3200m3Relative error P before and after blast furnace denoisingerrRespectively
For 0.0033 and 0.0027, root-mean-square error RMSE is respectively 0.0014 and 0.0011, is dropped respectively after two kinds of error value denoisings
Low 18% and 21%.Thus, it will be seen that Wavelet Denoising Method is passed through for same seat height stove utilization rate of carbon monoxide sample data
Processing error rate all obtains the larger (reduction of nearly 25%), so as to demonstrate the necessity of based Denoising.On the other hand, lead to
The error value crossed after comparing different blast furnace denoisings can obtain, 3200m3The relative error P of blast furnaceerrNumerical value (0.0027) compares 1100m3
Numerical value (0.0045) after denoising reduces 40%;For the latter, the former root-mean-square error RMSE compares (0.0028 He
0.0011) numerical value reduces 60%.Similar conclusion can also be compared by the error value before the two denoising and be obtained.
3200m is obtained at the same time it can also observe3Prediction error data value before large blast furnace denoising compares 1100m3After denoising
Error is also small.It can be illustrated by above-mentioned analysis, relative to medium and small blast furnace, the precision of prediction higher of large blast furnace.
In conclusion based on the blast furnace utilization rate of carbon monoxide Forecasting Methodology of chaos weighing first order local increment one
Determine to improve the accuracy that blast furnace utilization rate of carbon monoxide is predicted in degree, there is higher practical value in this way;
The embodiment of the present invention is described above in conjunction with attached drawing, but the invention is not limited in above-mentioned specific
Embodiment, above-mentioned specific embodiment is only schematical rather than restricted, those of ordinary skill in the art
Under the enlightenment of the present invention, present inventive concept and scope of the claimed protection are not being departed from, can also made several
It improves and deforms, these are belonged within the protection of the present invention.
Claims (9)
1. a kind of blast furnace CO utilization rates chaos weighing first order local prediction method, which is characterized in that include the following steps:
The time series η (co) of S1, the original blast furnace utilization rate of carbon monoxide of inputi, i=1,2 ..., N, wherein N are one oxygen of blast furnace
Change the length of carbon utilisation rate sample time-series;
S2, blast furnace one is determined using auto-relativity function method to the time series of original blast furnace utilization rate of carbon monoxide inputted in S1
Aoxidize carbon utilisation rate time series phase space insertion time τ;
S3, one oxygen of blast furnace is determined using correlation integral method to the time series of original blast furnace utilization rate of carbon monoxide inputted in S1
Change carbon utilisation rate time series phase space Embedded dimensions m;
S4, using in step S2 and S3 determine insertion time two parameters of τ and Embedded dimensions m, using delay coordinate method to height
Stove utilization rate of carbon monoxide time series carries out phase space reconfiguration;
S5, denoising is carried out to the time series phase space reconstruction obtained in step S4 using Wavelet noise-eliminating method;
S6, according to the blast furnace utilization rate of carbon monoxide time series phase space after denoising in step S5, choose phase space reconstruction most
The latter phase point is as prediction central point ZM, choose and prediction central point ZMDistance is siPoint be adjacent phase space point ZMi, determine
Adjacent phase space point ZMiWeights Qi;
S7, the adjacent phase space point Z obtained according to step S6MiWeights Qi, establish the weighting of blast furnace utilization rate of carbon monoxide chaos
Local increment, and calculate the coefficient matrix in first-order linear model of fit;
S8, according in step S7 determine optimal coefficient matrix, obtain blast furnace utilization rate of carbon monoxide predicted value.
2. according to claim 1 be based on blast furnace CO utilization rate chaos weighing first order local prediction methods, which is characterized in that
Determine that the embedded time is realized by following steps using auto-relativity function method in step S2:
S21, the auto-correlation function for establishing blast furnace utilization rate of carbon monoxide time series
S22, the auto-correlation function expression formula according to S21, work as Cττ values when value is intended to 0 can be identified as blast furnace carbon monoxide
The insertion time τ of utilization rate phase space reconstruction.
3. the blast furnace CO usage forecast methods according to claim 2 based on chaos weighing first order partial model, special
Sign is that the correlation integral method in step S3 determines that embedded dimension is realized by following steps:
S31, the number M for obtaining blast furnace utilization rate of carbon monoxide phase space reconstruction midpoint;
S32, the correlation integral functional equation for establishing blast furnace utilization rate of carbon monoxide sequential
Wherein, dijRepresent the Euclidean distance between arbitrary two vectors point in blast furnace utilization rate of carbon monoxide phase space reconstruction, r is phase
Optional distance in space,For Heaviside functions;
Correlation integral functional equation in S33, the numerical value and S32 of the insertion time τ that are determined according to step S2, obtains one oxygen of blast furnace
Change the calculation expression of carbon utilisation rate correlation dimension D:
S34, it can be calculated by the correlation dimension in step S33, the lnC of blast furnace sample space(co)m(r) with the change of the slope of lnr
When rate is less than preset threshold value k, m is smallest embedding dimension degree, so as to obtain the reconstruct of blast furnace utilization rate of carbon monoxide time series
Phase space Embedded dimensions m.
4. according to claim 1 be based on blast furnace CO utilization rate chaos weighing first order local prediction methods, which is characterized in that
Phase space reconfiguration is carried out to blast furnace utilization rate of carbon monoxide time series using delay coordinate method in step S4, by walking as follows
Suddenly it obtains:
S41:Input utilization rate of carbon monoxide phase space reconstruction insertion time τ and Embedded dimensions m;
S42:Phase space reconfiguration is carried out using delay coordinate method, the phase space after reconstruct is specially:
Wherein, N is the number of utilization rate of carbon monoxide original sample time series, and M=N- (m-1) τ is utilization rate of carbon monoxide
The vectorial number of time series embedding phase space.
5. according to claim 1 be based on blast furnace CO utilization rate chaos weighing first order local prediction methods, which is characterized in that
Weights Q in the step S6iSpecifically calculating process is:
It is calculated by equation below and determines adjacent phase space point ZMiWeights Qi;
Wherein, h is constant, usually takes the number that 1, p is consecutive points phase space point, siFor adjacent phase space point ZMiTo pre- measured center
Point ZMDistance, sminFor siIn minimum value.
6. according to claim 1 be based on blast furnace CO utilization rate chaos weighing first order local prediction methods, which is characterized in that
The step S7 the specific steps are:
S71:According to the weights Q of adjacent space point obtained in step S6i, establish blast furnace utilization rate of carbon monoxide chaos single order and add
Local increment is weighed, establishes adjacent phase space point ZMiWith kth step evolutionary phase spatial point ZMi+kBetween model;
ZMi+k=cke+gkZMi, i=1,2 ..., N
S72:Using following weighted least-squares method linear equation, weighted least-squares fitting linear function T and kth are established
Walk evolutionary phase spatial point ZMi+kBetween model;
Wherein,It is vector ZMiJ-th of element;
S73:Local derviation is asked to calculate simultaneously on the weighted least-squares method linear equation both sides obtained in step S72, determined most
Excellent linear fit coefficient matrix ckAnd gkValue.
7. according to claim 1 be based on blast furnace CO utilization rate chaos weighing first order local prediction methods, which is characterized in that
Blast furnace utilization rate of carbon monoxide predicted value in step S7 can be iterated to calculate by following formula to be obtained:
ZM+k=cke+gkZM
Wherein, e=(1,1 ..., 1)TIt is p dimensional vectors, ZM+kBlast furnace utilization rate of carbon monoxide predicted value is walked for kth.
8. according to claim 1 be based on blast furnace CO utilization rate chaos weighing first order local prediction methods, which is characterized in that
It further includes:The blast furnace utilization rate of carbon monoxide predicted value obtained by step S8, calculates the prediction of model as follows
Precision;
Obtain blast furnace utilization rate of carbon monoxide predicted valueBlast furnace utilization rate of carbon monoxide actual value z (m), prediction step number k;
According to formulaWithIt calculates opposite
Error (P based on chaos weighing first order partial modelerr) and root-mean-square error (RMSE);
Relative error (the P calculated according to two above formulaerr) and root-mean-square error (RMSE) come weigh based on chaos weight
The precision of prediction of single order partial model.
9. one kind is based on blast furnace CO utilization rate chaos weighing first order local prediction systems, which is characterized in that including:
Just enter initialization data information module, for inputting the time series η (co) of original blast furnace utilization rate of carbon monoxidei, i=
1,2 ..., N, wherein N are the length of blast furnace utilization rate of carbon monoxide sample time-series;
Embedded time module is calculated, for the time of the original blast furnace utilization rate of carbon monoxide of initialization data information module input
Sequence determines that blast furnace utilization rate of carbon monoxide time series phase space reconstruction is embedded in time τ using auto-relativity function method;
Embedded dimensions module is calculated, correlation integral is used for the time series of the original blast furnace utilization rate of carbon monoxide to input
Method determines blast furnace utilization rate of carbon monoxide time series phase space reconstruction Embedded dimensions m;
Phase space reconfiguration module, for carrying out phase space weight to blast furnace utilization rate of carbon monoxide time series using delay coordinate method
Structure;
Wavelet Denoising Method module, for using Wavelet noise-eliminating method to the utilization rate of carbon monoxide time series after phase space reconfiguration into
Row denoising;
Weight computing module, according to the blast furnace utilization rate of carbon monoxide time series phase space after Wavelet Denoising Method module denoising, really
Surely prediction central point ZM, find adjacent phase space point Z corresponding with the central pointMi, determine adjacent phase space point ZMiWeights Qi;
Model coefficient matrix acquisition module, for the adjacent phase space point Z obtained according to previous stepMiWeights Qi, establish blast furnace
Utilization rate of carbon monoxide chaos weighing first order local increment, and calculate the coefficient matrix in first-order linear model of fit;
Predicted value acquisition module according to determining optimal coefficient matrix, obtains blast furnace utilization rate of carbon monoxide predicted value.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711483980.5A CN108171381B (en) | 2017-12-29 | 2017-12-29 | Chaotic weighted first-order local prediction method and system for blast furnace CO utilization rate |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711483980.5A CN108171381B (en) | 2017-12-29 | 2017-12-29 | Chaotic weighted first-order local prediction method and system for blast furnace CO utilization rate |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108171381A true CN108171381A (en) | 2018-06-15 |
CN108171381B CN108171381B (en) | 2022-05-06 |
Family
ID=62516246
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711483980.5A Active CN108171381B (en) | 2017-12-29 | 2017-12-29 | Chaotic weighted first-order local prediction method and system for blast furnace CO utilization rate |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108171381B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110647717A (en) * | 2019-09-19 | 2020-01-03 | 中国水利水电科学研究院 | Hydrological model parameter optimization method based on correlation dimension |
CN112241174A (en) * | 2020-11-03 | 2021-01-19 | 山东理工大学 | Static obstacle avoidance method and system for security robot |
CN112347087A (en) * | 2020-09-30 | 2021-02-09 | 中国铁路上海局集团有限公司 | Quality control method for wind speed observation data along high-speed rail |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101505268A (en) * | 2009-03-12 | 2009-08-12 | 上海交通大学 | Public congestion path detecting method based on weighing first order local area process |
CN106845447A (en) * | 2017-02-19 | 2017-06-13 | 辽宁工程技术大学 | A kind of face gas concentration prediction method for early warning |
-
2017
- 2017-12-29 CN CN201711483980.5A patent/CN108171381B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101505268A (en) * | 2009-03-12 | 2009-08-12 | 上海交通大学 | Public congestion path detecting method based on weighing first order local area process |
CN106845447A (en) * | 2017-02-19 | 2017-06-13 | 辽宁工程技术大学 | A kind of face gas concentration prediction method for early warning |
Non-Patent Citations (3)
Title |
---|
DENGFENG XIAO ET AL: "The chaotic characteristic of the carbon-monoxide utilization ratio in the blast furnace", 《ISA TRANSACTIONS》 * |
张思梅: "基于多重分形理论的河流氨氮时空演变分析", 《安徽建筑工业学院学报(自然科学版)》 * |
郜传厚: "高炉铁水Si 含量的修正混沌加权一阶局部预报", 《物理学报》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110647717A (en) * | 2019-09-19 | 2020-01-03 | 中国水利水电科学研究院 | Hydrological model parameter optimization method based on correlation dimension |
CN110647717B (en) * | 2019-09-19 | 2020-05-05 | 中国水利水电科学研究院 | Hydrological model parameter optimization method based on correlation dimension |
CN112347087A (en) * | 2020-09-30 | 2021-02-09 | 中国铁路上海局集团有限公司 | Quality control method for wind speed observation data along high-speed rail |
CN112347087B (en) * | 2020-09-30 | 2024-05-03 | 中国铁路上海局集团有限公司 | Quality control method for high-speed rail wind speed observation data |
CN112241174A (en) * | 2020-11-03 | 2021-01-19 | 山东理工大学 | Static obstacle avoidance method and system for security robot |
Also Published As
Publication number | Publication date |
---|---|
CN108171381B (en) | 2022-05-06 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109659933B (en) | Electric energy quality prediction method for power distribution network with distributed power supply based on deep learning model | |
Zhao et al. | A statistical approach for interval forecasting of the electricity price | |
CN107885951B (en) | A kind of Time series hydrological forecasting method based on built-up pattern | |
CN107526927B (en) | Blast furnace molten iron quality online robust soft measurement method | |
CN101520652B (en) | Method for evaluating service reliability of numerical control equipment | |
CN114611844B (en) | Method and system for determining alloy addition amount in converter tapping process | |
CN107292344B (en) | Robot real-time control method based on environment interaction | |
CN107967803A (en) | Traffic congestion Forecasting Methodology based on multi-source data and variable-weight combined forecasting model | |
Omane-Adjepong et al. | Determining the better approach for short-term forecasting of ghana’s inflation: Seasonal ARIMA Vs holt-winters | |
CN105701559A (en) | Short-term load prediction method based on time sequence | |
CN108171381A (en) | A kind of blast furnace CO utilization rates chaos weighing first order local prediction method and system | |
CN110309608B (en) | Blast furnace molten iron silicon content forecasting method aiming at time lag uncertain information | |
CN104573857A (en) | Power grid load rate prediction method based on intelligent algorithm optimization and combination | |
CN103268519A (en) | Electric power system short-term load forecast method and device based on improved Lyapunov exponent | |
CN106529185B (en) | A kind of combination forecasting method and system of ancient building displacement | |
CN103853939A (en) | Combined forecasting method for monthly load of power system based on social economic factor influence | |
CN103885867B (en) | Online evaluation method of performance of analog circuit | |
CN103971289A (en) | Mine drawing data processing method and device | |
CN114548509A (en) | Multi-type load joint prediction method and system for multi-energy system | |
CN110837915A (en) | Low-voltage load point prediction and probability prediction method for power system based on hybrid integrated deep learning | |
CN115759445A (en) | Machine learning and cloud model-based classified flood random forecasting method | |
CN106295877B (en) | Method for predicting electric energy consumption of smart power grid | |
CN108830405B (en) | Real-time power load prediction system and method based on multi-index dynamic matching | |
CN111259340B (en) | Saturation load prediction method based on logistic regression | |
CN115423146A (en) | Self-adaptive runoff forecasting method based on multi-factor nearest neighbor sampling regression and support vector machine |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
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 |