Background technology
In the actual production process of coal-burning power plant, due to technology or economically, many important skills
Art parameter and economic parameters directly cannot be measured with conventional sensor, have a strong impact on coal-burning power plant's economy fortune
OK.Unburned carbon in flue dust is an important indicator of coal-fired boiler in thermal power plant efficiency of combustion, and it reflects
The size of coal-fired incomplete combustion loss.At present, unburned carbon in flue dust online test method has calcination to lose
Weight method and microwave detection method.
It is that " flying dust and slag can right thing mensuration sides for China electric power industry standard that calcination method weight-loss method measures technology
Method " and " proximate analysis of coal " in correlation technique, when the ash sample containing uncompleted burned carbon regulation
High temperature under after calcination, due in ash sample residual carbon made the quality of ash sample occur in that damage by the afterburnt
Lose, utilize the phosphorus content that the loss on ignition of ash sample calculates in ash sample as foundation,
The quality (%) of phosphorus content=[after before calcination, ash sample adds quality (g) calcination of crucible, ash sample adds crucible
Quality (g)]/[before calcination, ash sample adds the quality (g) of quality (g) the front crucible of receipts ash of crucible].
Fig. 1 shows the apparatus structure block diagram using calcination loss method to measure unburned carbon in flue dust.Such as Fig. 1 institute
Showing, the main working process of calcination loss method is to be collected by vibrator and sampler by the ash sample in flue
In the crucible of measuring unit, elevating mechanism crucible is positioned over rotary-tray, then by measuring unit
The actuator in portion will be equipped with crucible feeding firing device (e.g., furnace heater) of ash sample and carries out high temperature
Calcination, electronic balance measures the weight signal before receiving ash, after receipts ash and after calcination, control unit pair in real time
The weight signal received calculates, it is thus achieved that the phosphorus content of flying dust is the most enterprising at the display screen of control unit
Row display, wherein can use hand control box to control control unit and calculate or show.After calcination
Ash sample is drained back in flue by ash exhauster and the vacuum generator of system.
But there are more following problems in applying at the scene in calcination loss method:
(1) problem of sampling.Device sample rate is slow, so sample time is long, ash sample cooling is very fast,
The most stifled ash.
(2) microwave-oven-heating problem.First (power industry standard " flies can not to reach laboratory heating-up temperature
Ash and slag can right thing assay methods " in proposition should carry out calcination at 810 ± 10 DEG C, slow ash wants 1 little
Time, fast ash wants 30 minutes), calcination temperature, time can not reach requirement, and owing to microwave heating is special
Property and crucible contain the minimum requirements restriction of ash and ash sample thickness gone beyond the scope, thus can only be in crucible
The complete calcination of ash sample outer surface, and inside is former ash sample.Secondly, it is impossible to first dry by code moisture (because of
By difference variation factor in flue, flying dust is it sometimes appear that dew condensation phenomenon), survey volatile quantity, institute in calcination
So that weightless delta data to have moisture changing factor.
(3) weigh problem.Electronic balance is arranged in in-site measurement cabinet and weighs in real time, and boiler is transported
Row causes electronic balance to vibrate always, it is impossible to ensure normal certainty of measurement.
(4) mechanical problem.Such device typically has extremely complex machinery and motor, laughs somebody to scorn
Mechanical breakdown, electrical fault and crucible station mistake, often occur in using that crucible is broken, in rotatable parts
Etc. various mechanical breakdowns, due to equipment fault can relate to the machinery of complexity, Electrical connections,
Gas circuit, ash road system, even if there being special messenger to safeguard, it is also difficult to solve complicated problem.
(5) problem time lag.Device sample rate is slow and it needs to machine operation, weigh, calcination,
Make to detect cycle time long.
So, the on-line measuring device of calcination loss method have in using at the scene detection lag time length,
Ash sample can not grill thoroughly, it is inaccurate to weigh, stifled ash, defect that failure rate of machinery is high, and maintenance requirement is high, difficult
Safeguard.
The principle of microwave detection method is: because fly ash combustible material Main Ingredients and Appearance is the dielectric constant of carbon and carbon,
Microwave test unit may utilize the microwave signal of fixed frequency emitted energy homeostasis, and in flying dust, combustible contains
Measuring the highest, the effect absorbing microwave energy is the strongest.
The system of microwave detection method can use unpowered fly ash sampler, is automatically collected by the ash sample in flue
In the measurement pipe of microwave test device, ash level controller automatic discrimination collect the height of ash position.Work as receipts
When collection arrives enough ash samples, system carries out microwave resonance measurement to unburned carbon in flue dust.Control device and open electricity
Magnet valve accesses compressed air and purges the ash sample analyzed, and according to program setting or manually arranges, and flying dust can
Blow back flue with automatic sampled pipeline or send into receipts ash container, in order to chemical analysis is chemically examined.
The material composition complexity owing to the frequency characteristic of different material is different, in flying dust, and master during coal type change
If mineral change makes microwave detection method can not detect unburned carbon in flue dust change when coal type change.
Summary of the invention
It is an object of the invention to provide a kind of flying marking measuring method, for realizing unburned carbon in flue dust
On-line measurement, improves safety and the economy of unit operation.
To achieve these goals, the present invention provides a kind of flying marking measuring method, and the method includes:
Historical data values according to unburned carbon in flue dust and the auxiliary variable value corresponding to described historical data values are set up
Historical data sample set;First close with current working is chosen in described historical data sample set
Sample set;According to the auxiliary variable value in described first sample set and flying marking value matching for auxiliary
The weight coefficient of variable;And the auxiliary variable value and described weight coefficient according to current working calculates described
The unburned carbon in flue dust of current working.
Preferably, described in described historical data sample set, first sample close with current working is chosen
This collection includes: the auxiliary variable value calculating described current working is each with described historical data sample set
The correlation coefficient of the auxiliary variable value of sample;And choose more than corresponding to the correlation coefficient of first threshold
Sample forms described first sample set.
Preferably, Pearson product-moment correlation coefficient method is used to calculate described correlation coefficient.
Preferably, described method also includes: described first sample set is carried out principal component analysis to be wrapped
The second sample set containing main auxiliary variable;And come weighting system described in matching according to described second sample set
Number.
Preferably, weight coefficient described in least square fitting is used.
Preferably, described auxiliary variable includes: the coal-supplying amount of each feeder, flue gas oxygen content, smoke evacuation
Temperature, main steam pressure, main distillation amount, main steam temperature, total blast volume, economizer inlet flow rate value,
Economizer inlet pressure, economizer exit pressure, generation load, coal characteristic, First air gross pressure,
Secondary air register aperture, after-flame throttle opening, secondary wind stagnation pressure and burner hearth differential pressure, air preheater outlet temperature
Degree, burner pivot angle.
Preferably, described method also includes: update described historical data sample set.
Preferably, the set of described renewal described historical data sample includes: in described historical sample set
Determine multiple center of a sample;In emerging sample with the plurality of center of a sample, the distance of each is big
In the case of the second predetermined value, described emerging sample is added to described historical data sample collection
Close.
Preferably, the set of described renewal described historical data sample farther includes: by described emerging
Sample as a new center of a sample to update the plurality of center of a sample;And it is many after merging renewal
In individual center of a sample, distance each other is less than the Liang Ge center of a sample of the 3rd threshold value.
Preferably, the set of described renewal described historical data sample farther includes: after described renewal
In the case of distance each other is all higher than the 3rd threshold value in multiple center of a sample, delete going through after updating
With the distance reckling of each of the plurality of center of a sample in history data sample set.
By technique scheme, historical data sample set based on unburned carbon in flue dust and current working
Auxiliary variable value calculates the unburned carbon in flue dust of current working, and scheme is simple without complicated device, and
And safety and the economy of unit operation can be improved.
Other features and advantages of the present invention will be described in detail in detailed description of the invention part subsequently.
Detailed description of the invention
Below in conjunction with accompanying drawing, the detailed description of the invention of the present invention is described in detail.It should be appreciated that
Detailed description of the invention described herein is merely to illustrate and explains the present invention, is not limited to this
Bright.
Fig. 2 shows the flow chart of flying marking measuring method provided by the present invention in an embodiment.
As in figure 2 it is shown, the present invention provides a kind of flying marking measuring method, the method includes: according to flying dust
The historical data values of phosphorus content and the auxiliary variable value corresponding to described historical data values set up historical data
Sample set (step S10);Described historical data sample set is chosen close with current working
First sample set (step S20);According to the auxiliary variable value in described first sample set and flying marking
Value matching is for the weight coefficient (step S30) of auxiliary variable;And the auxiliary according to current working
Variate-value and described weight coefficient calculate the unburned carbon in flue dust (step S40) of described current working.
Hereinafter will the most specifically introduce flying marking measuring method provided by the present invention.
Step S10: according to the historical data values of unburned carbon in flue dust and auxiliary corresponding to described historical data values
Variate-value is helped to set up historical data sample set.
Wherein, between the multiple variable of unburned carbon in flue dust and scene, there is relation, in order to reflect that extraction flying dust contains
The multi-aspect information of carbon amounts, chooses the auxiliary variable big with unburned carbon in flue dust dependency here.This auxiliary becomes
Amount may include that the coal-supplying amount of each feeder (is assumed there are five feeders, then should be included feeder
A coal-supplying amount, feeder B coal-supplying amount, feeder C coal-supplying amount, feeder D coal-supplying amount, feeder E
Coal-supplying amount), flue gas oxygen content, exhaust gas temperature, main steam pressure, main distillation amount, main steam temperature,
Total blast volume, economizer inlet flow rate value, economizer inlet pressure, economizer exit pressure, generation load,
Coal characteristic, First air gross pressure, secondary air register aperture, after-flame throttle opening, secondary wind stagnation pressure and stove
Thorax differential pressure, air preheater outlet temperature, burner pivot angle etc..In the historical sample set set up
Each sample in comprise the value of above-mentioned auxiliary variable and the data value of the unburned carbon in flue dust of correspondence.
Step S20: choose first sample close with current working in described historical data sample set
Collection.
In this step, the auxiliary variable value that can calculate current working is every with historical data sample set
The correlation coefficient of the auxiliary variable value of one sample, correlation coefficient is the highest, illustrates that the two dependency is the highest, institute
Corresponding historical sample is got over close with current working.Here Pearson product-moment correlation coefficient method can be used
Calculate above-mentioned correlation coefficient.
If the auxiliary variable value of current working is expressed as X1=(x1, x2……xn), wherein x1, x2……
xnRepresenting the value of different auxiliary variables in current working, the auxiliary variable value of historical sample is Y1=(y1,
y2……yn), wherein y1, y2……ynRepresent the value of different auxiliary variables, then skin in a certain historical sample
The computing formula of you inferior product moment correlation coefficient r is:
In formula
SxxFor variable xiTo its averageSum of square of deviations, SyyFor variable yiTo its averageDeviation
Quadratic sum, SxyFor xi、yiSum of square of deviations.R be known as vector X1 to Y1 (or vector Y1
To X1) individual event correlation coefficient, also referred to as correlation coefficient.
The span of correlation coefficient r is-1≤r≤1, and the absolute value of correlation coefficient r is closer to 1, vectorial
Degree of correlation between X1 and Y1 is the highest, the absolute value of correlation coefficient r closer to 0, vector X1 and
Degree of correlation between Y1 is the lowest, chooses more than first threshold in the correlation coefficient calculated
Sample corresponding to correlation coefficient forms the first above-mentioned sample set.Here, first threshold can be according to tool
Depending on body situation, do not make specific restriction.
Step 30: according to the auxiliary variable value in the first sample set and flying marking value matching for auxiliary
The weight coefficient of variable.Preferably, main constituent can be carried out for the first sample set obtained here to divide
Analysis obtains comprising the second sample set of main auxiliary variable, according to the main auxiliary in this second sample set
Variate-value and flying marking value matching are for the weight coefficient of main auxiliary variable.
Can use PCA that the first sample set obtained is carried out main one-tenth in one embodiment
Analyze to obtain the second sample set.
PCA is the statistical method of a kind of dimensionality reduction, and it is by means of an orthogonal transformation, by its point
The former random vector that amount is relevant changes into the incoherent new random vector of its component, and this shows as on algebraically
The covariance matrix of former random vector is transformed into diagonal form battle array, is geometrically showing as former coordinate system transformation
The orthogonal coordinate system of Cheng Xin, is allowed to point to sample point and spreads p the orthogonal direction opened most, then to multidimensional
Variable system carries out dimension-reduction treatment, makes it to be converted into low-dimensional variable system with a higher precision, then
By constructing suitable cost function, further low-dimensional system is changed into unidimensional system.
Assuming that X is the data matrix of a n × m, each of which arranges corresponding to a variable, every a line
Corresponding to a sample in the first sample set.Matrix X can be decomposed into the apposition sum of m vector,
I.e.
In formula, ti∈RnIt is referred to as score vector, pi∈RmIt is referred to as load vector.The score vector of X is also named
Do the pivot of X.Can be write as lower column matrix formation
X=TPT (3)
Wherein T=(t1,t2,…,tn) it is referred to as score matrix, P=(p1,p2,…,pm) it is referred to as matrix of loadings.
Also it is orthogonal between each score vector, i.e. for any i and j, as i ≠ j, meets titj=0.
Also being orthogonal between each load vector, the length of the most each load vector is 1, i.e.
When there is a certain degree of linear correlation when between the vector in matrix X, the change of data matrix X will
On the direction of major embodiment several loads vector up front, data matrix X is rearmost several negative
Projection on lotus vector will be the least, and they are mainly due to measuring what noise caused.Thus can say
Matrix X is write as following formula after carrying out pivot decomposition
In formula, E is error matrix, represents X at pk+1To pmDeng the change in load vector direction.Due to by mistake
E, mainly due to measuring what noise caused, is neglected and often plays removing measurement noise by difference matrix E
Effect, without causing the significantly sacrificing of useful information in data.Thus data X can approximate earth's surface
It is shown as
The most just complete to be carried out main one-tenth for what the first sample set obtained carried out calculating formula (7)
Analyze.
Because front k the master of X represents the most changes in X data, therefore can be with before X
K pivot replaces those to be originally inputted variable carrying out regression analysis, the most just obtain following pivot and return
Return model
In formula, matrix Y represents the second sample set after principal component analysis,
B=(b1 b2 … bk)TFor principal component regression model parameter.Available least square method passes through following formula
Calculate and obtain B
Then model parameter θ of input variable is
Model parameter θ is calculated weight coefficient.The weight coefficient calculated is used to be multiplied by each
The auxiliary variable value of the current working corresponding to weight coefficient just can calculate the flying dust of current working and contain
Carbon amounts.
Further, in order to reflect operating mode situation comprehensively, when new sample occurs, need more new historical
Data sample set.First in historical data sample set, l (l >=1) individual center of a sample is determined, meter
Calculate emerging sample and the distance of each in l center of a sample, when each distance is all higher than threshold value δ
Time, then emerging sample is joined in historical data sample set, using this emerging sample as
One new center of a sample, center of a sample's number is l+1.Calculate l+1 center of a sample each other
Distance, if the distance of two samples is less than threshold value σ, then merges this Liang Ge center of a sample.As
Really the distance of two samples is all higher than threshold value σ, calculates each sample and l+1 sample in historical data base
The distance at center, rejects the sample that distance is minimum, thus ensures that the sample number in historical data base is protected
Hold constant.
In actual use, unburned carbon in flue dust computational methods provided by the present invention can be by supporting C language
The PLC (controller as serial in the S7 of Siemens) of speech programming realizes, logical
Cross Ethernet or hard wired form reads field data (auxiliary variable value), and result of calculation is shown
In the watch-dog of operations staff.
The preferred embodiment of the present invention is described in detail above in association with accompanying drawing, but, the present invention does not limit
Detail in above-mentioned embodiment, in the technology concept of the present invention, can be to the present invention
Technical scheme carry out multiple simple variant, these simple variant belong to protection scope of the present invention.
It is further to note that each the concrete technology described in above-mentioned detailed description of the invention is special
Levy, in the case of reconcilable, can be combined by any suitable means, in order to avoid need not
The repetition wanted, various possible compound modes are illustrated by the present invention the most separately.
Additionally, combination in any can also be carried out between the various different embodiment of the present invention, as long as its
Without prejudice to the thought of the present invention, it should be considered as content disclosed in this invention equally.