CN107831651A - Coal dust station boiler SCR and SNCR denitration system cooperative control method - Google Patents

Coal dust station boiler SCR and SNCR denitration system cooperative control method Download PDF

Info

Publication number
CN107831651A
CN107831651A CN201710945847.0A CN201710945847A CN107831651A CN 107831651 A CN107831651 A CN 107831651A CN 201710945847 A CN201710945847 A CN 201710945847A CN 107831651 A CN107831651 A CN 107831651A
Authority
CN
China
Prior art keywords
control
moment
data
scr
optimization
Prior art date
Application number
CN201710945847.0A
Other languages
Chinese (zh)
Inventor
刘永红
彭钢
徐欣航
张洪涛
高志存
殷喆
Original Assignee
国网河北能源技术服务有限公司
国家电网公司
国网河北节能服务有限公司
国网河北省电力公司电力科学研究院
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 国网河北能源技术服务有限公司, 国家电网公司, 国网河北节能服务有限公司, 国网河北省电力公司电力科学研究院 filed Critical 国网河北能源技术服务有限公司
Priority to CN201710945847.0A priority Critical patent/CN107831651A/en
Publication of CN107831651A publication Critical patent/CN107831651A/en

Links

Classifications

    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05BCONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/04Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
    • G05B13/042Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance

Abstract

The invention discloses coal dust station boiler SCR and SNCR denitration system cooperative control method, including combustion control and denitration spray ammonia control optimization;The combustion control includes DATA REASONING and acquisition, data cleansing, establishes steady state data storehouse, establishes rule base;Denitration spray ammonia control optimization includes the optimization of control structure and realizes the algorithm of denitration spray ammonia control optimization.Solves the problem of discharge numerical value of thermal NO x in thermal power plant's combustion process is high, and fluctuation is big;Solve the problems, such as that SCR device is too high due to the escaping of ammonia brought for ammonia regulation;Solves blockage problem caused by air preheater the escaping of ammonia;Improve SCR region, denitration outlet NOx regulation stability;With stronger popularization and application ability.

Description

Coal dust station boiler SCR and SNCR denitration system cooperative control method

Technical field

Present invention relates particularly to coal dust station boiler SCR and SNCR denitration system cooperative control method.

Background technology

Newly issued according to State Ministry of Environmental Protection《Thermal power plant's nitrogen oxides technological policy for treatment》It has been shown that, low-NO_x combustion technology should As the one preferred technique of coal-burning power plant's nitrogen oxides control.The emphasis of control is the fire coal of 200,000 kilowatts and the above in the whole country All Thermal generation units and cogeneration of heat and power in generating set and cogeneration units and atmosphere pollution priority control region Unit.

Domestic and international research institution has all carried out substantial amounts of research work to the burning optimization in power station, it is proposed that data mining is advised Then technology and the product such as foundation, running optimizatin technology, intelligent control.

Data mining technology has obtained fast development since proposition, and wherein U.S. Zapton companies are answered in its engineering Include the application example for being based partially on KDD technologies in:Data mining technology is used to reduce the cigarette in Lime Factory's discharge flue gas Dust content and the diagnosis for petrochemical equipment.The TonyOgilvie and B.W.Hogg of Queens universities, which are proposed, utilizes data The model that information service is established under the conditions of power plant's full working scope full operation, a system, Neng Goujing are developed on the basis of model Really recognize operation conditions and the operating condition that Operating Guideline is monitored with correcting can be provided for operations staff and deviate.With foreign countries Compare, a little later, the current country has many colleges and universities and R&D institution to be working on data mining for the domestic research to data mining Basic theory and application study, these units include Tsing-Hua University, Institute of Computing Technology, CAS, research institute of air force the 3rd, Naval equipment demonstration center etc..Wherein, Beijing Research Institute of System Engineering is carried out to application of the blur method in Knowledge Discovery Compared with in-depth study;Peking University is also carrying out the research to data cubical algebra;Fudan University, Zhejiang University, Chinese section The units such as skill university, Chinese Academy of Sciences's Institute of Mathematics, the Central China University of Science and Technology have been carried out the optimization to association rules mining algorithm and changed Make;The Knowledge Discovery of unstructured data is inquired into, have studied to the units such as Nanjing University, Sichuan Union University, Shanghai Communications University And web data is excavated, North China Electric Power University and Southeast China University are carried out in fired power generating unit using correlation rule to service data Extraction, and emerge a collection of outstanding achievement in research.But the precision that data mining technology is applied to simulation model is tested Card also without ripe achievement in research.

In terms of optimizing operation, foreign well-known has American Electric Power research institute (EPRI) in terms of power station performance diagnogtics and monitoring Leadership is constantly in, develops unit operation energy management system, thermal efficiency research software, but it has black-box model The characteristics of.North China Electric Power University is developed based on heat balance method of moral boiler combustion monitoring system, and each coal is measured using heat balance method of The coal powder density of powder pipeline, and wind speed measuring point and coal burning caloricity on-line monitoring system and oxygen content in exhaust gas and smoke evacuation in installing pipe Temperature monitoring measuring point, wind speed in each coal pipe and coal powder density and boiler thermal output can be monitored, instruct operation people Member's operation maintains the equilibrium of each tube cell coal amount, and the system does not provide optimal scheme and can not risen to reducing NOx emission To directive function.

In terms of controlling optimisation technique, external main control optimisation technique is mainly derived from CONTROLSOFT, west gate at present The control optimization company of the specialty such as son, GE, ASPENTECH is engaged in the public into business men, wherein controlsoft of DCS system The software kernels of department are in the on-line amending with PID control, and the design for control strategy and scheme is real with domestic power plant project Border greatly differs from each other.The PROFI of Siemens Company has very with its familiarity to unit to the tactful compactness of Control of Power Plant Strong advantage.It exists and is difficult to safeguard simultaneously, it is not easy to secondary development, the problem of cost is high, domestic few again after 2014 There is achievement.It is engaged in most of debugging public affairs for based on Xi'an hot working research institute and each province's DianKeYuan of control optimizing research in the country Department, such enterprise's field experience is enriched, but service quality varies with each individual, and commercialization feature is undesirable, and quality can not be protected Card.

The content of the invention

The purpose of the present invention is to overcome defect present in above-mentioned prior art and provide one kind and be applied to coal dust power station pot Stove SCR and SNCR denitration system cooperative control method, solve the discharge numerical value height of thermal NO x in thermal power plant's combustion process, The problem of fluctuation is big;Solve the problems, such as that SCR device is too high due to the escaping of ammonia brought for ammonia regulation;Solves air preheater ammonia Blockage problem caused by escape;Improve SCR region, denitration outlet NOx regulation stability;It can apply to all configurations SCR adds the unit of low nitrogen burning mode, has stronger popularization and application ability.

SCR:Selective Catalytic Reduction, as SCR technology.

SNCR:Selective no-Catalytic Reduction, as SNCR technology.

The purpose of the present invention is achieved through the following technical solutions:

Coal dust station boiler SCR and SNCR denitration system cooperative control method, including combustion control and denitration spray ammonia control Optimization;The burning includes DATA REASONING and acquisition, data cleansing with SNCR optimal controls, establishes steady state data storehouse, foundation rule Then storehouse, prediction module;Denitration spray ammonia control optimization includes the optimization of control structure and realizes the algorithm of denitration spray ammonia control optimization;

In the burning and SNCR optimal controls, prediction module to SNCR entrances NOx by setting and predicting that NOx is exported, finally Realize the output control of urea;

The optimization of the control structure comprises the following steps:

The big optimization of NOx deviations at SCR outlets and chimney neat stress:By the self-calibration cycle of air preheater outlet NOx analysis meters It is set as 2 hours, sampling probe carries out a blowback for every 2 hours and swept, and the demarcation signal of analysis meter is introduced into DCS, demarcation is opened Blowback is carried out after beginning to sampling probe to sweep 120 seconds;

The optimization delayed be present in SCR inlet NOx measurement signals:Entrance NOx measurements are there is larger purely retarded, from entrance Mechanism is set out caused by NOx, and SCR inlet NOx is influenceed by boiler combustion, realizes that entrance NOx is predicted by data fusion;

It is DMC algorithms that the denitration spray ammonia, which controls the algorithm of optimization, is comprised the following steps:

(1)Step response

Determine the sample frequency of step response and react the sampled point of total response process, i=1,2 ... ..., T ... ...;T is to adopt The inverse of sample frequency;At this time point of kT, work as incrementSystem output is represented by after working(1)

(1)

Refer to the prediction output of the controlled volume at N+1 moment;It is the prediction output of initial time controlled volume;

According to the principle of PREDICTIVE CONTROL,Always it is expected to be finally intended to zero;It can makeTended in the form of index In zero;Attenuation factor it is expected in setting, and α is a constant from 0 to 1;So its controlling increment can be expressed as from the K moment

ΔU(k)=[1,α,α2,…α M-1]TΔu(k)=AΔu(k) (2)

It can be represented with method for simplifying at t=kT moment in increment processedIn the presence of,System is in following P The individual moment prediction output by(3)Represent

(3)

Prediction output after elapsed time sequence,It is the prediction output of initial time controlled volume;A is rank Jump response matrix, from testing inspection;

(2)Rolling optimization

In sampling instant t=kT optimality criterion J(k), it is set as such as formula(4)It is shown,

(4)

In order that controlled quentity controlled variable, and controlled volume is as close possible to and making control output to control in rational oscillating region, control The final of system output is desired for Δu(k)=0;Formula(4)In,,For weight coefficient, P and M are referred to as optimization time domain length and control Time domain length processed;

In the case where not considering input and output constraint, at t=kT moment,It is that oneself knows, J (k) is taken most SmallExtreme value necessary condition can be passed throughTry to achieve

Δu(k)= (A u T QA u +A α T λA α T)-1 A u T Q[Yr(k+1)-A 0 U(k-1)-he(k)] (5)

Make (A u T QA u +A α T λA α T)-1=D;Δu(k) be the optimum control increment sequence solved at t=kT moment, i.e. open loop is optimal Solution;It can be made by the optimization of this volumeA u=AA αOne column vector, D are the calculating that a pure values ask reciprocal, so as to reduce meter Calculate step;Q is the matrix of coefficient;

(3)Feedback compensation

Implement Δ first at t=kT momentu(k) in first control action

Δu(k)= (A u T QA u +A α T λA α T)-1 A u T Q[Yr(k+1)-A 0 U(k-1)-he(k)] (5)

(6)

U (k) is that the control at k moment exports;

U (k-1) is the control output at the upper a moment at k moment;

Coefficient matrix dTSuch as formula(7)It is shown,

(7)

Current error such as formula(8)It is shown,

(8)

y(k+1)For the numerical value of the actual controlled volume at k+1 moment;

For the estimate of the actual controlled volume at k+1 moment;

To current error e (k+1) weight coefficient, h, (i=1,2 ..., N) is obtained

(9)

Formula(9)InExisted by the system that t=(k+1) T moment is predicted after error correction t=(k+i)T(I=1 ..., N)The output at moment;For the estimate of the actual controlled volume at k moment;For error Updating vector, wherein,;After calibratedEach component in, remaining item of Except First Item are t=(k+ respectively 1) the T moment there is noTo exporting in t=(k+2) T when being acted on etc. following controlling increment ..., the prediction at (k+N) T moment Value, when they can be used as t=(k+1) TPreceding N-1 component, i.e.,, I=1 ..., N-1 andIn last component;That is prediction of t=(k+1) the T moment to i=(k+1+N) T outputs, can ByIt is next approximate, i.e.,, the expression of above-mentioned relation availability vector form

(10)

WhereinFor shift matrix;

At t=(k+1) T moment, have, repeat the new prediction optimization of t=kT moment progress, and then rolling optimization Realize whole control process;Wherein, to entrance NO in unit load and total blast volume selective binding prediction moduleXPrediction make multiply Method, when total blast volume measurement is credible, manually using total blast volume as multiplication factor;When total blast volume measurement is problematic, then switch to Stressor, realize that the control of denitration spray ammonia optimizes with this.

As a further improvement on the present invention, the DATA REASONING and acquisition are specially to obtain data, bag from monoblock First air, Secondary Air pressure are included, First air, Secondary Air speed, burner hearth oxygen amount, coal pulverizer combination start stopping mode, burner Pivot angle or baffle opening, after-flame air quantity, the supplemental characteristic of carbon monoxide and oxynitrides carry out data fusion.

As a further improvement on the present invention, the data cleansing is specially SIS systems with according to certain frequency gathered data, The rinsing ratio set as 2, can by the use of 1min as criterion, if it is bad point not occur the value in the time to be considered as it, It is then bad point and rejects this one piece of data;And clean, obtain separated one by one again within the time after the bad point starts Data time section, that is, complete the data cleansing process to this parameter value;The wherein threshold value of measuring point and rinsing ratio can be changed.

As a further improvement on the present invention, it is described establish steady state data storehouse operation it is as follows:In the period of data cleansing In the range of, data search is carried out, finds the period for meeting correlated variables threshold range and correlated variables constraints, is steady The state period;For load variation, it is desirable to the probability upper and lower 5% reaches 90%, in the range of upper and lower 10% data reach 95% with On.

As a further improvement on the present invention, it is unit load, fire coal, air pressure balance condition according to the decision method of stable state Difference different operating points is divided into different stable states;The division of load setting can use 2 the percent of total load to make For a steady state point;Coal is divided into high nitrogen element content, middle nitrogen element content, low nitrogen element content coal;Air pressure has passed through pressure Power deviation and two aspects of rate of change of pressure divergence have carried out criterion;Mapped by operating mode and complete performance analysis, and to newest Caused steady result is divided, the steady-state distribution looked under certain operating mode, and the operating mode in operating mode list can show correlation Stability recording.

As a further improvement on the present invention, stable state search is actively carried out, when choosing stable state after Steady-state Parameters determination Between Duan Houhui jump to supplemental characteristic trend page and record the trend, and the steady state time section chosen is set and preserved, mistake herein Characteristic value and preservation of all system variables in each steady state time section are calculated in journey;The steady state time section that editor's current search arrives And result set, without manual intervention, all interventions stay in the completion of data cleansing stage.

As a further improvement on the present invention, it is described establish rule base operation it is as follows:In above-mentioned data mining results On the basis of carry out Rule Extraction;After whole extracting rules, some front and back pieces can cause conflict in rule tree, regular chain In related front and back pieces all separated from rule tree;This Result is saved in stable state number by extracting rule after terminating According in storehouse, the modification of Result can not be carried out after preservation again.

As a further improvement on the present invention, specified rule chain specifies the regular data after operating mode, lists rule tool The front and back pieces parameter numerical digit of body, is analyzed from set state to combination property or is divided from firing optimization to set state Analysis.

As a further improvement on the present invention, the nitrogen oxygen of the SCR region of the whole unit of Collaborative Control is carried out by SNCR and SCR Compound, the peak of the very big numerical value of SCR inlet NOx and instantaneous rate of change is mainly stabilized by SNCR effect, as shown in figure 4, Know NOx numerical value in advance by prediction, can targetedly act the variable quantity of the concentration for the NOx for removing 90%.

As a further improvement on the present invention, after the controlled quentity controlled variable for formulating parameters, the most suitable operation control of artificial selection Scheme processed is entered in DCS by PLC, completes the real-time control to unit operation;Unit exists more when running under some operating mode The optimal section of individual performance synthesis, make unit operation close to nearest optimized operation section rapidly in real time, complete in rule base Rule operation, certain operation is carried out to " adjustable controllable factor ", ensures that unit can operate in optimum state as far as possible, completes To the combustion control of boiler.

The present invention employs DMC control technologies in SCR region, improves the process of the matrix inversion in DMC, is changed into numerical value and asks It is reciprocal;Reduce operand, improve the stability of unit.

NOx constructive ways are mainly heating power NOx, quick NOx and fuel NO_x.The discharged nitrous oxides and pot of boiler The efficiency of furnace is had a strong impact on by combustion conditions, using the method for controlling combustion of optimization, it is ensured that boiler low NOx drainage and Obtain low unburned carbon in flue dust, high boiler efficiency.

Practical experience shows, does not transform combustion system, by firing optimization method with regard to that can obtain 10%-25% or so NOx Reducing effect, and operating cost is very cheap.

By the burning optimization technology based on data mining, the rule base of a boiler operatiopn is determined, can be to unit All load sections, optimizing is carried out under different coal pulverizers, the machine supplying powder method of operation, determines a series of best operating points of unit, The optimum-curve for operation of unit can be determined by the operating point of magnanimity;After obtaining optimum-curve for operation, ammonia is sprayed with denitration Control optimisation technique ensures its technical feasibility.

SCR gas denitrifying technologies, reducing agent ammonia is sprayed into flue gas, by the NOx conversion in flue gas be N using catalyst2 And water.After minimum discharge standard starts execution, catalyst layer and air preheater blockage problem are increasingly serious.Therefore SCR region outlet NOx Concentration stability contorting, the operation of air preheater long-term safety, denitration efficiency, catalyst life, the escaping of ammonia and the economy of denitration operation Property etc. be intended in denitrification process balance control.From formation NOXAspect on say, running optimizatin based on data mining will It is to solve NOXThe problem of few discharge, stable emissions.

The cooperative control method of the present invention, according to the statistics of middle Electricity Federation relevant technical project, to more than 300MW machines Group due to air preheater block and the annual single unit of water developing fee of reduction can save 200,000 yuan;Subtract for air preheater differential pressure Economic well-being of workers and staff that is small and bringing can convert into electricity 165000KW;600,000 yuan of the economic benefit brought is saved by spray ammonia;Solve The problem of thermal NO x discharge numerical value is high in thermal power plant's combustion process, and fluctuation is big;SCR device is solved due to being adjusted for ammonia Save the problem of the escaping of ammonia brought is too high;Solves blockage problem caused by air preheater the escaping of ammonia;Improve SCR region, denitrification apparatus Export NOx regulation stability;With stronger popularization and application ability.

Brief description of the drawings

Fig. 1 is the data flow structured flowchart in data mining of the present invention;

Fig. 2 is data and production procedure structured flowchart needed for the combustion control optimization of unit in the present invention;

Fig. 3 is that DMC algorithms realize block diagram;

Fig. 4 is the control flow block diagram of prediction module;

Fig. 5 is DMC control optimization block diagrams.

Embodiment

Present disclosure is further explained and illustrated with specific embodiment below in conjunction with the accompanying drawings.

A kind of coal dust station boiler SCR of the present invention and SNCR denitration system cooperative control method, including burning optimization control System and denitration spray ammonia control optimization, concrete operations control method are as follows:

Burning and SNCR optimal controls

As depicted in figs. 1 and 2, DATA REASONING and acquisition:Obtained from monoblock in data, including pressure, speed, fluid machinery Portion's flowing, oxygen at furnace exit, burning structure parameter, the supplemental characteristic of coal pulverizer operation combination.

Data cleansing:Specially SIS systems are with according to certain frequency gathered data, the rinsing ratio set is 2, Ke Yiyong 1min is as criterion, if it is bad point not occur the value in the time to be considered as it, for bad point and rejects this one piece of data; And cleaned again within the time after the bad point starts, obtain data time section separated one by one, that is, complete to join this The data cleansing process of numerical value;The wherein threshold value of measuring point and rinsing ratio can be changed.

Establish steady state data storehouse:In the range of the period of data cleansing, data search is carried out, finds and meets correlated variables The period of threshold range and correlated variables constraints, as steady state time section;For load variation, it is desirable to upper and lower 5% Probability reaches 90%, and data reach more than 95% in the range of upper and lower 10%.

Decision method according to stable state be unit load, fire coal, air pressure balance condition it is different by different operating mode dot-dash It is divided into different stable states;The division of load setting can use percent the 2 of total load to be used as a steady state point;Coal is divided into High nitrogen element content, middle nitrogen element content, low nitrogen element content coal;Air pressure has passed through the rate of change of pressure divergence and pressure divergence Two aspects have carried out criterion;Mapped by operating mode and complete performance analysis, and newest caused steady result is divided, looked for The steady-state distribution sought under certain operating mode, the operating mode in operating mode list can show the stability recording of correlation.

Stable state search is actively carried out, is become after Steady-state Parameters determination choosing steady state time Duan Houhui to jump to supplemental characteristic Gesture page simultaneously records the trend, and the steady state time section chosen is set and preserved, and calculates all system variables in the process each Characteristic value and preservation in steady state time section;Editor's current search the steady state time section and result set that arrive, without manual intervention, All interventions stay in the completion of data cleansing stage.

Rule Extraction is carried out on the basis of above-mentioned data mining results;After whole extracting rules, some front and back pieces meetings Cause conflict in rule tree, the related front and back pieces in regular chain is all separated from rule tree;Extracting rule knot This Result is saved in steady state data storehouse after beam, the modification of Result can not be carried out after preservation again.

Specified rule chain specifies the regular data after operating mode, lists the specific front and back pieces parameter numerical digit of the rule, slave Group state is analyzed to combination property or analyzed from firing optimization to set state.

The nitrogen oxides of the SCR region of the whole unit of Collaborative Control is carried out by SNCR and SCR, it is main by SNCR effect It is the peak for stabilizing the very big numerical value of SCR inlet NOx and instantaneous rate of change, knows NOx numerical value in advance by prediction, there can be pin Action to property removes the variable quantity of 90% NOx concentration.

DMC control technologies are employed in the control of SCR region, and improve the process of the matrix inversion in DMC, are changed into several Value asks reciprocal;The stability of unit is improved on the basis of operand is reduced.

After the controlled quentity controlled variable for formulating parameters, the most suitable operation control program of artificial selection is entered in DCS by PLC, Complete the real-time control to unit operation;There is the optimal section of multiple performance synthesises in unit, real when running under some operating mode When make unit operation close to nearest optimized operation section rapidly, the rule operation in rule base is completed, to " adjustable controllable Factor " carries out certain operation, ensures that unit can operate in optimum state as far as possible, completes the combustion control to boiler.

The Optimization Platform based on big data can not only reduce NOX's from the angle of boiler combustion as seen in Figure 2 Discharge, while NOX predicted value is also outputed, provide the advanced argument on controlling for SNCR input urea liquid.Such as Fig. 4 The control flow block diagram of shown prediction module, the forecast function of prediction module are integrated each big data in abovementioned steps, And by being exported to SCR inlet NOx settings, prediction NOx, finally realize the output control of urea;It is achieved in burning and SNCR Optimal control.

Now by with illustrate how to obtain data exemplified by Xinhua XDPS control systems:

XDPS tabulating system includes two large divisions's function:Data Collection and data representational role.Data collection program is used The concept of Open Database Connectivity (ODBC), the data of collection can be write in the data source that ODBc is defined by the user.So User can both use XDPS historical datas to reproduce software transfer, statistics, printing reports, can also use general data base administration Instrument reproduces the data collected, and improves the opening of system.It is the application program based on EXCEL that XDPS tabulated datas, which reproduce, EXCEL functions are taken full advantage of, make form configuration simple and convenient.XDPS forms include preiodic type form, flip-over type form, recollected Data type form, SOE types form, event mode form, self-defined schedule report.

Preiodic type form refers to the form formed within the regular hour, such as Times, class report, daily paper, monthly magazine, cycle The minimum time unit of type form is 1 hour.Data type and species in form are determined by form configuration file.Schedule report Data source configuration step is as follows:

1. open " ODBC " data source manager;

2. click " addition ' item, " establishment source of new data " window will be ejected;

3. selecting " MieorsotfAeeesSDriver (* .mdb) ", click " completion ";

4. input data parent name " tbadat ", describing " schedule report database ", clicking " establishment ", " new database " window will be ejected Mouthful;

5. input data library name " batd.atmbd ", Path selection " f:Data tbadat ", click OK;

6. system ejects " database F:Idsat atbad.atmdb create successfully " prompt window.

The step of using form configurator (c.sfxls) configuration cycle form, is as follows:

1. open form configurator

2. configuration cycle form

3. in click menu column " Xl) PS " selections " storage preiodic type form collection procedure configuration file ", input configuration file Tbaero.nii complete trails filenames.System will change batero.ini automatically, preserve configuration.

So far, XDPS reporting systems can gather boiler combustion system from control network in real time by advance definition format Service data, and be stored on local hard drive;It is EXCEL by data table transmition using " export " function of ACCESS databases Electrical form.

Denitration spray ammonia control optimization

1)Control structure

(1)The big corrective measure of NOx deviations at SCR outlets and chimney neat stress:Due to the distribution of SCR reactor outlets NOx concentration not , under-represented, air preheater exhaust pass flue gas flow field is relatively stable, and measurement air preheater exit NOx concentration participates in denitration spray ammonia Self-regulated is practicable.NOx deviations at NOx averages and neat stress are exported to reduce A, B side air preheater, air preheater is exported The self-calibration cycle set of NOx analysis meters is 2 hours, so as to ensure the accuracy of its measurement;Blocked to reduce sampling probe Phenomenon occurs, and sampling probe carries out a blowback for every 2 hours and swept;Due to being both needed to keep data when instrument self-calibration and probe purging Constant, air preheater outlet NOx actual values keep numerical value to have relatively large deviation when being purged with demarcation, are unfavorable for denitration self-regulated Adjustment;Therefore the time that instrumented data keeps constant need to be reduced as far as possible, the demarcation signal of analysis meter is introduced into DCS, demarcation starts Carry out blowback to sampling probe afterwards to sweep 120 seconds, so as to reduce the generation of instrument measurement error and stopping state of popping one's head in.

(2)SCR inlet NOx measurement signals, which exist, delays corrective measure, overcomes this problem main from SCR inlet NOx Influence factor is started with, and is found out and is reflected that the feedforward amount of entrance NOx changes carrys out look-ahead entrance NOx changes in advance.Actual motion passes through Test and show, in the case of Secondary Air total blast volume is constant, when SOFA burnout degrees(Separated OFA, separate burnout degree)Air door is opened During degree increase, primary zone oxygen concentration can reduce, and now, inlet NOx concentration is on a declining curve;Secondary Air air distribution is to entrance NOx Concentration is influenceed, and upper, middle and lower-ranking secondary air register aperture is reduced, and can reduce primary zone excess air coefficient, increases burnout degree, Axial Researched of Air Staging Combustion Burning Pulverized Coal in stove is realized, now, inlet NOx concentration is on a declining curve;There is larger for entrance NOx measurements Purely retarded, from mechanism caused by entrance NOx, SCR inlet NOx is influenceed by boiler combustion, realized by data fusion Entrance NOx is predicted.

With following DMC algorithms(See Fig. 3)Realize the control of denitration spray ammonia:

(1)Step response

Determine the sample frequency of step response and react the sampled point of total response process, i=1,2 ... ..., T ... ...;T is to adopt The inverse of sample frequency;At this time point of kT, work as incrementSystem output is represented by after working(1)

(1)

Refer to the prediction output of the controlled volume at N+1 moment;It is the prediction output of initial time controlled volume;

According to the principle of PREDICTIVE CONTROL,Always it is expected to be finally intended to zero;It can makeTended in the form of index In zero;Attenuation factor it is expected in setting, and α is a constant from 0 to 1;So its controlling increment can be expressed as from the K moment

ΔU(k)=[1,α,α2,…α M-1]TΔu(k)=AΔu(k) (2)

It can be represented with method for simplifying at t=kT moment in increment processedIn the presence of,System is in following P The individual moment prediction output by(3)Represent

(3)

Prediction output after elapsed time sequence,It is the prediction output of initial time controlled volume;A is step Response matrix, from testing inspection;

(2)Rolling optimization

In sampling instant t=kT optimality criterion J(k), it is set as such as formula(4)It is shown,

(4)

In order that controlled quentity controlled variable, and controlled volume is as close possible to and making control output to control in rational oscillating region, control The final of system output is desired for Δu(k)=0;Formula(4)In,,For weight coefficient, P and M are referred to as optimization time domain length and control Time domain length processed;

In the case where not considering input and output constraint, at t=kT moment,It is that oneself knows, J (k) is taken most SmallExtreme value necessary condition can be passed throughTry to achieve

Δu(k)= (A u T QA u +A α T λA α T)-1 A u T Q[Yr(k+1)-A 0 U(k-1)-he(k)] (5)

Make (A u T QA u +A α T λA α T)-1=D;Δu(k) be the optimum control increment sequence solved at t=kT moment, i.e. open loop is optimal Solution;It can be made by the optimization of this volumeA u=AA αOne column vector, D are the calculating that a pure values ask reciprocal, so as to reduce meter Calculate step;Q is the matrix of coefficient;

(3)Feedback compensation

Implement Δ first at t=kT momentu(k) in first control action

Δu(k)= (A u T QA u +A α T λA α T)-1 A u T Q[Yr(k+1)-A 0 U(k-1)-he(k)] (5)

(6)

U (k) is that the control at k moment exports;

U (k-1) is the control output at the upper a moment at k moment;

Coefficient matrix dTSuch as formula(7)It is shown,

(7)

Current error such as formula(8)It is shown,

(8)

y(k+1)For the numerical value of the actual controlled volume at k+1 moment;

For the estimate of the actual controlled volume at k+1 moment;

To current error e (k+1) weight coefficient, h, (i=1,2 ..., N) is obtained

(9)

Formula(9)InExisted by the system that t=(k+1) T moment is predicted after error correction t=(k+i)T(I=1 ..., N)The output at moment;For the estimate of the actual controlled volume at k moment;For error Updating vector, wherein,;After calibratedEach component in, remaining item of Except First Item are t=(k+ respectively 1) the T moment there is noTo exporting in t=(k+2) T when being acted on etc. following controlling increment ..., the prediction at (k+N) T moment Value, when they can be used as t=(k+1) TPreceding N-1 component, i.e.,, I=1 ..., N-1 andIn last component;That is prediction of t=(k+1) the T moment to i=(k+1+N) T outputs, can ByIt is next approximate, i.e.,, the expression of above-mentioned relation availability vector form

(10)

WhereinFor shift matrix;

At t=(k+1) T moment, have, repeat the new prediction optimization of t=kT moment progress, and then rolling optimization Realize whole control process;Wherein, to entrance NO in unit load and total blast volume selective binding prediction moduleXPrediction make multiply Method, when total blast volume measurement is credible, manually using total blast volume as multiplication factor;When total blast volume measurement is problematic, then switch to Stressor, realize that the control of denitration spray ammonia optimizes with this.

DMC control optimization block diagrams as shown in Figure 5, by above-mentioned DMC algorithms, to unit load and total blast volume selectivity With reference to entrance NO in prediction moduleXPrediction make multiplication, when total blast volume measurement it is credible when, manually using total blast volume as multiplication because Son;When total blast volume measurement is problematic, then stressor is switched to, realize that the control of denitration spray ammonia optimizes with this.

The present invention employs DMC control technologies in SCR region, improves the process of the matrix inversion in DMC, is changed into numerical value and asks It is reciprocal;Reduce operand, improve the stability of unit.

Claims (10)

1. a kind of coal dust station boiler SCR and SNCR denitration system cooperative control method, it is characterised in that including following control step Suddenly:
(1)Burning with SNCR optimal controls, including DATA REASONING and acquisition, data cleansing, establish steady state data storehouse, establish rule Storehouse and prediction module;
(2)Denitration spray ammonia control optimization, including control structure optimization and realize denitration spray ammonia control optimization algorithm.
2. coal dust station boiler SCR according to claim 1 and SNCR denitration system cooperative control method, its feature exist In the optimization of the control structure comprises the following steps:
The big optimization of NOx deviations at SCR outlets and chimney neat stress:By the self-calibration cycle of air preheater outlet NOx analysis meters It is set as 2 hours, sampling probe carries out a blowback for every 2 hours and swept, and the demarcation signal of analysis meter is introduced into DCS, demarcation is opened Blowback is carried out after beginning to sampling probe to sweep 120 seconds;
The optimization delayed be present in SCR inlet NOx measurement signals:Entrance NOx measurements are there is larger purely retarded, from entrance Mechanism is set out caused by NOx, and SCR inlet NOx is influenceed by boiler combustion, realizes that entrance NOx is predicted by data fusion;
The algorithm of the denitration spray ammonia control optimization comprises the following steps:
(1)Step response
Determine the sample frequency of step response and react the sampled point of total response process, i=1,2 ... ..., T ... ...;T is to adopt The inverse of sample frequency;At this time point of kT, system is exported and is represented by after increment works(1)
(1)
According to the principle of PREDICTIVE CONTROL,Always it is expected to be finally intended to zero;It can makeTended in the form of index In zero;Attenuation factor it is expected in setting, and α is a constant from 0 to 1;So its controlling increment can be expressed as from the K moment
ΔU(k)=[1,α,α2,…α M-1]TΔu(k)=AΔu(k) (2)
It can be represented with method for simplifying at t=kT moment in controlling incrementIn the presence of,System is not Come the prediction output at P moment by(3)Represent
(3)
Exported for the prediction after elapsed time sequence,It is the prediction output of initial time controlled volume;A is rank Jump response matrix, from testing inspection;
(2)Rolling optimization
In sampling instant t=kT optimality criterion J(k), it is set as such as formula(4)It is shown,
(4)
In order that controlled quentity controlled variable, and controlled volume is as close possible to and making control output to control in rational oscillating region, control The final of system output is desired for Δu(k)=0;
In the case where not considering input and output constraint, at t=kT moment,It is that oneself knows, J (k) is taken most SmallExtreme value necessary condition can be passed throughTry to achieve
Δu(k)= (A u T QA u +A α T λA α T)-1 A u T Q[Yr(k+1)-A 0 U(k-1)-he(k)] (5)
Make (A u T QA u +A α T λA α T)-1=D;The optimum control increment sequence that as t=kT moment solves, i.e. open loop are optimal Solution;It can be made by the optimization of this volumeA u=AA αOne column vector, D are the calculating that a pure values ask reciprocal, so as to reduce meter Calculate step;Q is the matrix of coefficient;
(3)Feedback compensation
First control action in implementing first at t=kT moment
Δu(k)= (A u T QA u +A α T λA α T)-1 A u T Q[Yr(k+1)-A 0 U(k-1)-he(k)] (5)
(6)
U (k) is that the control at k moment exports;
U (k-1) is the control output at the upper a moment at k moment;
Coefficient matrix dTSuch as formula(7)It is shown,
(7)
Current error e (k+1), such as formula(8)It is shown,
(8)
y(k+1)For the numerical value of the actual controlled volume at k+1 moment;
For the estimate of the actual controlled volume at k+1 moment;
To current error e (k+1) weight coefficient, h (i=1,2 ..., N) is obtained
(9)
Formula(9)InThe system predicted after error correction by t=(k+1) T moment is in t =(k+i)T(I=1 ..., N)The output at moment;For the estimate of the actual controlled volume at k moment;For error Updating vector, wherein,;After calibratedEach component in, remaining item of Except First Item are t=(k respectively + 1) the T moment there is noTo output in t=(k+2) T ... when being acted on etc. following controlling increment, (k+N) T moment it is pre- Measured value, when they can be used as t=(k+1) TPreceding N-1 component, i.e., i= 1 ..., N-1 andIn last component, i.e. prediction output of t=(k+1) the T moment to i=(k+1+N) T can be byIt is next approximate, i.e.,, the expression of above-mentioned relation availability vector form
(10)
WhereinFor shift matrix;
At t=(k+1) T moment, have, the new prediction optimization of t=kT moment progress is repeated, and then rolling optimization is real Now whole control process;
In the burning and SNCR optimal controls, prediction module to SNCR entrances NOx by setting and predicting that NOx is exported, finally Realize the output control of urea;
To entrance NO in unit load and total blast volume selective binding prediction moduleXPrediction make multiplication, when total blast volume measure it is credible When, manually using total blast volume as multiplication factor;When total blast volume measurement is problematic, then stressor is switched to, is realized with this de- The control optimization of nitre spray ammonia.
3. coal dust station boiler SCR according to claim 1 and SNCR denitration system cooperative control method, its feature exist In:The DATA REASONING and acquisition are specially to obtain data, including First air, Secondary Air pressure, First air, two from monoblock Secondary wind velocity, burner hearth oxygen amount, coal pulverizer combination start stopping mode, the pivot angle or baffle opening, after-flame air quantity, an oxygen of burner The supplemental characteristic for changing carbon and oxynitrides carries out data fusion.
4. coal dust station boiler SCR according to claim 1 and SNCR denitration system cooperative control method, its feature exist In:The data cleansing is specially SIS systems with according to certain frequency gathered data, the rinsing ratio set can use 1min as 2 As criterion, if it is bad point not occur the value in the time to be considered as it, for bad point and this one piece of data is rejected;And Cleaned again within the time after the bad point starts, obtain data time section separated one by one, that is, complete to this parameter value Data cleansing process;The wherein threshold value of measuring point and rinsing ratio can be changed.
5. coal dust station boiler SCR according to claim 4 and SNCR denitration system cooperative control method, its feature exist In:Decision method according to stable state is that different operating points is divided into by the difference of unit load, fire coal and air pressure balance condition Different stable states;The division of load setting can use 2 the percent of total load to be used as a steady state point;Coal is divided into height Nitrogen element content, middle nitrogen element content, low nitrogen element content coal;Air pressure has passed through the rate of change two of pressure divergence and pressure divergence Individual aspect has carried out criterion;Mapped by operating mode and complete performance analysis, and newest caused steady result is divided, looked for Steady-state distribution under certain operating mode, the operating mode in operating mode list can show the stability recording of correlation;Stable state search is actively carried out, Steady state data storehouse is choosing steady state time Duan Houhui to jump to supplemental characteristic trend page and is recording the trend after determining, and will choose Steady state time section set preserve, calculate characteristic value and guarantor of all system variables in each steady state time section in the process Deposit;The steady state time section and result set that editor's current search arrives, without manual intervention, all interventions stay in the data cleansing stage After the completion of.
6. coal dust station boiler SCR according to claim 5 and SNCR denitration system cooperative control method, its feature exist In carrying out Rule Extraction on the basis of above-mentioned data mining results;After whole extracting rules, some front and back pieces can be in rule Cause conflict in tree, the related front and back pieces in regular chain is all separated from rule tree;Extracting rule will after terminating This Result is saved in steady state data storehouse, can not carry out the modification of Result after preservation again.
7. coal dust station boiler SCR according to claim 6 and SNCR denitration system cooperative control method, its feature exist In:Specified rule chain specifies the regular data after operating mode, the specific front and back pieces parameter numerical digit of the rule is listed, from set state Analyzed to combination property or analyzed from firing optimization to set state.
8. coal dust station boiler SCR according to claim 7 and SNCR denitration system cooperative control method, its feature exist In:The nitrogen oxides of the SCR region of the whole unit of Collaborative Control is carried out by SNCR and SCR, by SNCR effect, stabilizes SCR The peak of the very big numerical value of entrance NOx and instantaneous rate of change, know NOx numerical value in advance by prediction, targetedly act simultaneously Remove the variable quantity of 90% NOx concentration.
9. coal dust station boiler SCR according to claim 8 and SNCR denitration system cooperative control method, its feature exist In:DMC control technologies are used in SCR region, improve the process of the matrix inversion in DMC, is changed into numerical value and asks reciprocal;Reduce computing Amount, improve set steady.
10. coal dust station boiler SCR according to claim 9 and SNCR denitration system cooperative control method, its feature exist In:After the controlled quentity controlled variable for formulating parameters, the most suitable operation control program of artificial selection is entered in DCS by PLC, completion pair The real-time control of unit operation;The optimal section of multiple performance synthesises be present, make machine in real time in unit when running under some operating mode Group operation is close to nearest optimized operation section rapidly, completes the rule operation in rule base, adjustable controllable factor is entered The certain operation of row, ensure that unit can operate in optimum state as far as possible, complete the combustion control to boiler.
CN201710945847.0A 2017-10-12 2017-10-12 Coal dust station boiler SCR and SNCR denitration system cooperative control method CN107831651A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710945847.0A CN107831651A (en) 2017-10-12 2017-10-12 Coal dust station boiler SCR and SNCR denitration system cooperative control method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710945847.0A CN107831651A (en) 2017-10-12 2017-10-12 Coal dust station boiler SCR and SNCR denitration system cooperative control method

Publications (1)

Publication Number Publication Date
CN107831651A true CN107831651A (en) 2018-03-23

Family

ID=61647820

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710945847.0A CN107831651A (en) 2017-10-12 2017-10-12 Coal dust station boiler SCR and SNCR denitration system cooperative control method

Country Status (1)

Country Link
CN (1) CN107831651A (en)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104296131A (en) * 2014-10-23 2015-01-21 东南大学 Multivariable cooperative control method for double-hearth circulating fluidized bed unit
CN104607042A (en) * 2014-12-26 2015-05-13 东南大学 Selective catalytic reduction (SCR) denitration system and method based on constraint predictive control
CN105629738A (en) * 2016-03-24 2016-06-01 内蒙古瑞特优化科技股份有限公司 SCR (Selective Catalytic Reduction) flue gas denitration system control method and apparatus
US20170167341A1 (en) * 2015-12-10 2017-06-15 General Electric Company System and method for emission control in power plants
CN107103176A (en) * 2017-01-17 2017-08-29 东南大学 Coal-burning boiler SCR catalyst life assessment method based on multisource information fusion technology

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104296131A (en) * 2014-10-23 2015-01-21 东南大学 Multivariable cooperative control method for double-hearth circulating fluidized bed unit
CN104607042A (en) * 2014-12-26 2015-05-13 东南大学 Selective catalytic reduction (SCR) denitration system and method based on constraint predictive control
US20170167341A1 (en) * 2015-12-10 2017-06-15 General Electric Company System and method for emission control in power plants
CN105629738A (en) * 2016-03-24 2016-06-01 内蒙古瑞特优化科技股份有限公司 SCR (Selective Catalytic Reduction) flue gas denitration system control method and apparatus
CN107103176A (en) * 2017-01-17 2017-08-29 东南大学 Coal-burning boiler SCR catalyst life assessment method based on multisource information fusion technology

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
杜中文: "火电机组脱硝系统预测控制应用研究", 《中国优秀硕士学位论文全文数据库 工程科技Ⅰ辑》 *

Similar Documents

Publication Publication Date Title
US8527072B2 (en) System and method for multi-unit optimization of boiler and SCR operations
Smrekar et al. Multi-step-ahead prediction of NOx emissions for a coal-based boiler
CN105137760B (en) A kind of denitration spray ammonia autocontrol method and system
Backreedy et al. Modelling pulverised coal combustion using a detailed coal combustion model
Ahmadi et al. Thermodynamic analysis and thermoeconomic optimization of a dual pressure combined cycle power plant with a supplementary firing unit
Liu et al. A dynamic model used for controller design of a coal fired once-through boiler-turbine unit
Havlena et al. Application of model predictive control to advanced combustion control
Iliyas et al. RBF neural network inferential sensor for process emission monitoring
Gu et al. Online adaptive least squares support vector machine and its application in utility boiler combustion optimization systems
CN102494714B (en) Synchronous reckoning method of utility boiler efficiency and coal heat value as well as ash content and moisture content
Fast et al. Development and multi-utility of an ANN model for an industrial gas turbine
CN103792933B (en) Utilize the method that analogue system is determining and tunes process characteristic parameter
US8352148B2 (en) System for controlling input profiles of combined cycle power generation system
EP1921280B1 (en) Systems and methods for multi-level optimizing control systems for boilers
CA2639197C (en) Dual model approach for boiler section cleanliness calculation
Bayón et al. The exact solution of the environmental/economic dispatch problem
CN103197549B (en) The hard measurement of Circulating Fluidized Bed Boiler sulfur dioxide in flue gas and optimal control method
CN101799170B (en) Method for correcting fuel calorific capacity of coal-fired boiler in real time
Luo et al. Operational planning optimization of multiple interconnected steam power plants considering environmental costs
CN102032590B (en) Boiler combustion optimizing control system and optimizing control method based on accurate measurement system
Hu et al. The cost of carbon capture and storage for coal-fired power plants in China
CN102734783B (en) Method for calibrating monitoring data parameters of heat-absorbing surfaces at each level of supercritical boiler of power station
CN103322547B (en) Boiler control and combustion optimization method
JP4270218B2 (en) Control device for control object having combustion device, and control device for plant having boiler
US20140316737A1 (en) Method for monitoring inside-boiler dynamic wall temperature of power plant boiler high-temperature piping system

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination