CN104763999A - Power plant pulverized coal boiler combustion performance online optimizing method and system - Google Patents
Power plant pulverized coal boiler combustion performance online optimizing method and system Download PDFInfo
- Publication number
- CN104763999A CN104763999A CN201510094539.2A CN201510094539A CN104763999A CN 104763999 A CN104763999 A CN 104763999A CN 201510094539 A CN201510094539 A CN 201510094539A CN 104763999 A CN104763999 A CN 104763999A
- Authority
- CN
- China
- Prior art keywords
- boiler
- data
- pulverized coal
- variable
- boiler combustion
- 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.)
- Pending
Links
Landscapes
- Regulation And Control Of Combustion (AREA)
- Control Of Steam Boilers And Waste-Gas Boilers (AREA)
Abstract
The invention provides a power plant pulverized coal boiler combustion performance online optimizing method and system. The method comprises the steps of receiving basic data of boilers in different load work conditions, which are acquired by a Distributed Control System (DCS) and an instrument, wherein the basic data is used for establishing non-linear mapping relationship between pulverized coal boiler adjustable input variables and output variables through a radial basis function network, and the non-linear mapping relationship is used as a boiler combustion mathematic model; obtaining an optimal input combination and value of boiler combustion system adjustable variables under the corresponding expectation coal consumption and NOX emission level. By the aid of the method and the system, optimal control can be performed on boiler operation engineering, relationships among all operation parameters of the boiler are coordinated, the safety, the economy and the reliability of the system are further improved, and the boiler combustion system comprehensive performance is improved comprehensively.
Description
Technical field
The present invention relates to boiler technology field, particularly relate to a kind of pulverized coal power boiler combustibility method for on-line optimization and system.
Background technology
It is a complicated physical and chemical process that station boiler runs, and relates to the ambits such as Combustion, hydrodynamics, thermodynamics.Any detection with boiler operatiopn relevant parameter, adjustment, the transformation of equipment, steam generator system optimization can be called, comprise DCS (Distributed Control System, the dcs) optimization of control logic, the design of Controlling model.From steam generator system optimisation technique angle, boiler combustion optimization technology can be divided three classes: the first kind, by the major parameter of on-line checkingi boiler operatiopn, instructs operations staff to regulate boiler operatiopn operating mode, and this kind of optimisation technique is at present at home in occupation of leading position.Equations of The Second Kind boiler combustion optimization technology, in equipment aspect, is adjusted by the burning optimization transformation of burner, heating surface etc. being realized to boiler.3rd class, boiler combustion optimization technology, on the basis of DCS, as the supervisory control system of boiler operatiopn, by adopting advanced control logic, control algolithm and artificial intelligence technology, realizes the performance optimization of boiler.The monoblock put into operation in recent years, pervasive device is in good condition, and controllability is strong, automatization level is high, should make full use of information technology on this basis and exploit potentialities, pass through optimal control, realize lean production and control, thus surmount original extensive management and control.Along with the progressively ripe of Dynamic matrix control and artificial intelligence technology with industrially successfully apply, the 3rd class optimisation technique obtains fast development.Above-mentioned three types of technology respectively has advantage and application in practice, but wherein three types of technology does not need to carry out any transformation to boiler plant, the service data of boiler can be made full use of, on the basis that DCS controls, by the application of advanced modeling, optimization, control technology, directly improve boiler operating efficiency, reduce nitrogen oxide (NOx) discharge, have that small investment, risk are little, the advantage of successful, thus become the optimisation technique of priority research and development.Investigation situation shows, there are the following problems to some extent for thermal power plant's pulverized-coal fired boiler:
1), boiler thermal output is on the low side, and net coal consumption rate remains high;
2), load and ature of coal changeable, the frequent off-target combustion position of boiler;
3) oxygen amount, is crossed higher;
4), NOx emission exceeds standard;
5), fly ash combustible material and large slag total amount of combustible higher;
6), boiler coke slagging scorification.
Above problem is all affecting the security of equipment, economy and reliability index in varying degrees.Therefore, the matter of utmost importance that the optimization of boiler combustion system overall performance just becomes the needs solution of Present Thermal Power factory how is realized.
Therefore, occurred various power boiler burning state-detection and complex optimal controlled strategy system, these control systems achieve good effect in respective application scenario, but still there are the following technical issues that need to address:
One, the use of historical data
Boiler regulated variable (input variable is contained in a large amount of historical datas that boiler longtime running produces, as, first and second air quantity, secondary air register aperture, after-flame throttle opening, burner pivot angle, coal-supplying amount, coal blending mode, boiler load etc.) and output variable (boiler coal consumption, NOx discharge capacity) between mathematical relationship.How to effectively utilize these mass datas and set up the Mathematical Modeling that can be used for optimizing, in prior art, do not provide feasible method.
Two, adaptive modeling
How in conjunction with following service data, upgrade in time model, with the change of adaptation condition and ature of coal, is also the key issue involved by boiler combustion system is optimized.The instability of as-fired coal kind, adds the impact of the factors such as boiler maintenance, dust stratification, slagging scorification, and make the boiler model mismatch set up on performance test data basis serious, the online adaptive correction utilizing up-to-date burning data to carry out model seems especially important.If can the Mathematical Modeling of online updating boiler combustion system, then coal varitation and boiler overhaul etc. can be regarded its exterior service condition as and ignored, and like this, the optimum results based on online updating model has certain adaptive capacity.
Three, the optimizing under Prescribed Properties is realized under how ensureing flameholding
After MATHEMATICAL MODEL OF COMBUSTION is set up, how to select suitable optimized algorithm, under constrained condition, find adjustable input combination and the optimal solution plan of corresponding optimum output valve, traditional optimized algorithm fails to provide concrete scheme.
In addition, the searching process of existing burning optimization method mostly adopts the solution procedure of the inversion model of forward direction Mathematical Modeling.There is two problems like this: the existence of a. boiler combustion process inversion model is difficult to guarantee, oppositely solve and may produce without separating or the situation of separating more.B. the change between optimization solution and currency is excessive, means that the input amplitude of accommodation of combustion system is excessive like this, likely threatens combustion stability and security.Combustion control software needs the stability to optimizing existence of solution and burning to take into full account, only under solution unconfined condition, performance objective optimization problem is far from being enough, and the burning optimization problem that actual needs solves often exists multiple constraint.
Summary of the invention
The present invention is intended to propose a kind of pulverized coal power boiler combustibility method for on-line optimization and system, can improve the security of system, economy and reliability further based on this on line optimization system.
First aspect, the invention provides a kind of pulverized coal power boiler combustibility method for on-line optimization, comprises the steps: step 1, and the boiler receiving DCS system and instrument and meter collection is in the basic data under different load operating mode; Step 2, according to described basic data, employing radial primary function network sets up the Nonlinear Mapping relation between the adjustable input variable of pulverized-coal fired boiler and output variable, and described Nonlinear Mapping relation is as the Mathematical Modeling of boiler combustion; Step 3, according to this Mathematical Modeling, obtains corresponding input combination and the optimal value of expecting boiler combustion system regulated variable under coal consumption and NOX emission level.
In above-mentioned pulverized coal power boiler combustibility method for on-line optimization, step 3 comprises further: adopt genetic algorithm, under described radial primary function network parameter, find the adjustable input combination of suitable boiler combustion system and optimal solution, make combustion process coal consumption minimum, meet the requirement of the NOx emission upper limit simultaneously.
In above-mentioned pulverized coal power boiler combustibility method for on-line optimization, genetic algorithm, when determining the adjustable input combination of boiler combustion system and optimal solution, if without separating or having multiple solution, then continues through self adaptation multidimensional search algorithm and determines each variable-value in input combination.
In above-mentioned pulverized coal power boiler combustibility method for on-line optimization, step 3 comprises further, and the historical data of boiler combustion is regularly incorporated kainogenesis data, and regular update, in order to train described radial primary function network, obtains the Mathematical Modeling optimized.
In above-mentioned pulverized coal power boiler combustibility method for on-line optimization, the historical data of boiler combustion is regularly incorporated kainogenesis data, from the close-by examples to those far off change according to the time, give the forgetting factor that data are specified, distance current time data far away to forget degree higher, it is 0 that current data then forgets degree.
Second aspect, present invention also offers a kind of pulverized coal power boiler combustibility on line optimization system, comprising: basic data acquisition module, Mathematical Modeling determination module and regulated variable input combination and optimal solution determination module.Wherein, basic data acquisition module is in the basic data under different load operating mode for the boiler receiving DCS system and instrument and meter collection; Mathematical Modeling determination module is used for according to described basic data, and employing radial primary function network sets up the Nonlinear Mapping relation between the adjustable input variable of pulverized-coal fired boiler and output variable, and the Nonlinear Mapping relation determined is as the Mathematical Modeling of boiler combustion; Regulated variable input combination and optimal solution determination module are used for according to this Mathematical Modeling, obtain correspondence and expect coal consumption and NO
xthe input combination of boiler combustion system regulated variable and optimal value under emission level.
In above-mentioned pulverized coal power boiler combustibility on line optimization system, regulated variable input combination and optimal solution determination module comprise the first optimizing unit, this the first optimizing unit is for adopting genetic algorithm, under described radial primary function network parameter, find the adjustable input combination of suitable boiler combustion system and optimal solution, make combustion process coal consumption minimum, meet the requirement of the NOx emission upper limit simultaneously.
In above-mentioned pulverized coal power boiler combustibility on line optimization system, regulated variable input combination and optimal solution determination module also comprise the second optimizing unit: the second optimizing unit for adopt the adjustable input of genetic algorithm determination boiler combustion system to combine and optimal solution time, if without separating or having multiple solution, then continue through self adaptation multidimensional search algorithm and determine each variable-value in input combination.
In above-mentioned pulverized coal power boiler combustibility on line optimization system, Mathematical Modeling determination module is further used for, and the historical data of boiler combustion is incorporated kainogenesis data, and regular update, in order to train described radial primary function network, obtains the Mathematical Modeling optimized.
In above-mentioned pulverized coal power boiler combustibility on line optimization system, the mode historical data of boiler combustion regularly being incorporated kainogenesis data is, from the close-by examples to those far off change according to the time, give the forgetting factor that data are specified, distance current time data far away to forget degree higher, it is 0 that current data then forgets degree.
The present invention adopts RBF (Radial Basis Function, RBF) network sets up the Nonlinear Mapping relation between the adjustable input variable of pulverized-coal fired boiler and output variable, solve multivariable nonlinearity fitting problems with this, open the new way solving boiler combustion optimization and control.Utilize statistics multi-variable nonlinear regression to screen the input variable set of minimum necessity, the complexity of simplified model, improve the extensive and predictive ability of model.The present invention utilizes this model to carry out predicting or oppositely solving, obtain the value of each input variable under corresponding expectation coal consumption and NOx emission level, thus control can be optimized to boiler operatiopn, coordinate the relation between each operational factor of boiler, the security of further raising system, economy and reliability, improve overall boiler performance comprehensively.
Accompanying drawing explanation
By reading hereafter detailed description of the preferred embodiment, various other advantage and benefit will become cheer and bright for those of ordinary skill in the art.Accompanying drawing only for illustrating the object of preferred embodiment, and does not think limitation of the present invention.And in whole accompanying drawing, represent identical parts by identical reference symbol.In the accompanying drawings:
Fig. 1 is the flow chart of steps of pulverized coal power boiler combustibility method for on-line optimization preferred embodiment of the present invention;
Fig. 2 is in pulverized coal power boiler combustibility method for on-line optimization of the present invention, sets up boiler steady-state model principle schematic with radial primary function network;
Fig. 3 is in pulverized coal power boiler combustibility method for on-line optimization of the present invention, the topological structure schematic diagram of radial primary function network;
Fig. 4 is in pulverized coal power boiler combustibility method for on-line optimization of the present invention, the flow chart of steps of genetic algorithm;
Fig. 5 is the principle of optimality block diagram of pulverized coal power boiler combustibility method for on-line optimization of the present invention;
Fig. 6 is the schematic diagram of the control station adopting pulverized coal power boiler combustibility method for on-line optimization of the present invention;
Fig. 7 is the schematic diagram from the boiler performance optimal way shown in another angle;
Fig. 8 is the global design flow chart that boiler performance is optimized;
Fig. 9 is the structured flowchart of pulverized coal power boiler combustibility on line optimization system preferred embodiment of the present invention.
Detailed description of the invention
Below with reference to accompanying drawings exemplary embodiment of the present disclosure is described in more detail.Although show exemplary embodiment of the present disclosure in accompanying drawing, however should be appreciated that can realize the disclosure in a variety of manners and not should limit by the embodiment set forth here.On the contrary, provide these embodiments to be in order to more thoroughly the disclosure can be understood, and complete for the scope of the present disclosure can be conveyed to those skilled in the art.
Below in conjunction with Fig. 1, the preferred embodiment of pulverized coal power boiler combustibility method for on-line optimization of the present invention is described.
With reference to Fig. 1, the present embodiment pulverized coal power boiler combustibility method for on-line optimization comprises the steps:
Step S110, the boiler receiving DCS (Distributed Control System, dcs) system and instrument and meter collection is in the basic data under different load operating mode.
Step S120, according to basic data, employing radial primary function network sets up the Nonlinear Mapping relation between the adjustable input variable of pulverized-coal fired boiler and output variable, and Nonlinear Mapping relation is as the Mathematical Modeling of boiler combustion.
The present embodiment adopts RBF (Radial Basis Function, RBF) network sets up the Nonlinear Mapping relation between the adjustable input variable of pulverized-coal fired boiler and output variable, solve multivariable nonlinearity fitting problems with this, open the new way solving boiler combustion optimization and control.Utilize statistics multi-variable nonlinear regression to screen the input variable set of minimum necessity, the complexity of simplified model, improve the extensive and predictive ability of model.This model is utilized to carry out predicting or oppositely solving, obtain the value of each input variable under corresponding expectation coal consumption and NOx emission level, thus optimization guidance is carried out to boiler operatiopn, the relation between each operational factor of coherent system, reaches the object comprehensively improving overall boiler performance indications.
Step S130, according to this Mathematical Modeling, adopt genetic algorithm, under radial primary function network parameter, find the adjustable input combination of suitable boiler combustion system and optimal solution, make combustion process coal consumption minimum, meet NOx (nitrogen oxide) simultaneously and discharge upper limit requirement, obtain corresponding expectation coal consumption and NO
xthe input combination of boiler combustion system regulated variable and optimal solution under emission level; Genetic algorithm, when determining the adjustable input combination of boiler combustion system and optimal solution, if without separating or having multiple solution, then continues through self adaptation multidimensional search algorithm and determines each variable-value in input combination.
The present embodiment searching process comprehensively adopts genetic algorithm and self adaptation multi-dimensional search method to realize.Under the radial primary function network parameter that existing training completes, find the adjustable input combination of suitable boiler combustion system and optimal solution, make combustion process coal consumption minimum, meet the requirement of the NOx emission upper limit simultaneously.Genetic algorithm can ask for the globally optimal solution under Prescribed Properties, ensures that adjustable input solution all meets the restriction of its minimax, ensures boiler operatiopn security.If there is the situation without solution or multiple solution, then pass through the input solution in self adaptation multi-dimensional search method substep solving-optimizing direction, ensure that the combustion process under regulating by a small margin is stablized.
Step S140, regularly incorporates kainogenesis data by the historical data of boiler combustion, and regular update, in order to train radial primary function network, obtains the Mathematical Modeling optimized, and then returns and performs step S130.Wherein, it is from the close-by examples to those far off change according to the time further that the historical data of boiler combustion is regularly incorporated kainogenesis data, gives the data forgetting factor of specifying, the data that distance current time is far away to forget degree higher, it is 0 that current data then forgets degree.
Specifically, when being subject to the affecting of the working conditions change such as ature of coal, environment temperature, boiler plant overhaul at boiler combustion process, kainogenesis data are rationally incorporated, the original combustion system model of regular update, to adapt to contingent working conditions change at any time in history training data.The situation that the present embodiment from the close-by examples to those far off changed according to the time, gives data certain forgetting factor, the data that distance current time is far away to forget degree higher, current data then " without forgeing " incorporates training dataset.Like this, can ensure that the network after online updating can outside operating mode after " in time " Adaptive change, realize rolling dynamic optimization.
In the process of boiler combustion system founding mathematical models, other mathematical method can also be used, as the object using linear regression model (LRM), ARTOICAL NEURAL NETWORK MODEL, supporting vector machine model etc. all can realize modeling.But on the generalization ability of modeling accuracy and network and the present invention may difference to some extent.
After obtaining boiler burnup system model, the Optimization Solution process that model exports target can adopt multiple optimized algorithm optimizing, as: traditional climbing method, gradient method, steepest descent method, ant group algorithm, particle swarm optimization algorithm etc.But in efficiency of algorithm, engineering realizability, find that the overall situation has the method difference to some extent adopted with the present embodiment in the ability of solving etc. most.
Can find out, above-mentioned S120, S130 and S140 relate generally to three partial contents:
The first, boiler combustion system input, output data processing and boiler combustion system modeling;
Three, genetic algorithm and self adaptation multidimensional search algorithm combine and carry out boiler combustion system optimization;
Three, the Mathematical Modeling (history, current data) of boiler combustion is upgraded according to the large data of history and current data.Now be described as follows:
1) effect of pulverized coal power boiler combustibility method for on-line optimization depends on the precision of boiler Mathematical Modeling to a great extent, how to design by experiment, the input that acquisition modeling needs, output data, and on the basis of experimental data, set up the boiler Mathematical Modeling that can reflect truth, be the problem that first design of boiler performance optimization system needs to solve.
The present embodiment formulates testing program, Data acquisition and issuance plan in detail, variable bounds is obtained with the scheme of simplifying, by the minimal set of statistical calculations and correlation analysis input variable, process variable and output variable, with the disaster avoiding boiler modeling process to be absorbed in dimension, radial primary function network evolutionary training is degenerated, matching cannot input really, export mapping.
DCS gathers the various instrumented datas of steam generator system, can provide data and the information of its operation conditions of reflection.Boiler expert, on the basis to system profound understanding, sets up the economy of each service data information and boiler, quantitative relationship between safety and the performance indications such as reliable.Utilize statistics multi-variable nonlinear regression and radial primary function network technology, numerical relationship model between input variable and boiler performance index can be set up.Then this model is utilized to carry out predicting or oppositely solving, obtain the value of each input variable under corresponding expectation coal consumption and NOx emission level, thus optimization guidance is carried out to boiler operatiopn, the relation between each operational factor of coherent system, reaches the object comprehensively improving overall boiler performance.
In the present embodiment, the Nonlinear Mapping relation between input variable and intermediate variable carrys out modeling by RBF (Radial Basis Function, RBF) network, solves multivariable nonlinearity fitting problems with this.Radial primary function network module has the ability (self-organizing of stronger adaptation complex environment and multi objective control requirement, self study, self adaptation), trained by effective data, set up boiler operatiopn Mathematical Modeling, and in running, control being optimized to operation variable, the running status making steam generator system be tending towards best, can adapt to change and the disturbance of inside and outside portion environment.Boiler steady-state model process is set up as shown in Figure 2 with radial primary function network.
Structure is seen, RBF network belongs to multilayer feedforward neural network.It is a kind of three-layer forward networks, and input layer is made up of signal source node; The second layer is hidden layer, and the number of hidden unit is determined by described problem, and the transforming function transformation function Φ of hidden unit is to central point radial symmetric and the non-negative nonlinear function of decay; Third layer is output layer, and it makes response to input pattern.Radial basis function network topological structure as shown in Figure 3.Wherein, (x
1, x
2... x
n) (or be called regulated variable for input variable, independent in scope adjustment can be limit in safety), comprise the baffle openings such as First air, Secondary Air, burnout degree, surrounding air, racemization wind, burner nozzle angle, the instantaneous coal-supplying amount of feeder, unit load, coal qualities test index etc.(y
1, y
2... y
k) be output variable (also known as optimization aim amount), comprise net coal consumption rate (g/kWh) and NOx (mg/m
3).Activation primitive (transforming function transformation function of the hidden unit) Φ of hidden layer, is called kernel function, choosing for kernel function, has multiple choices (e.g., Gaussian function, nucleus vestibularis triangularis function, two exponential kernel functions etc.).Present patent application adopts gaussian kernel function.The basic thought of RBF network is: with RBF as hidden unit " base ", form hidden layer space, hidden layer converts input vector, by the pattern of low-dimensional input data transformation in higher dimensional space, the linearly inseparable problem in lower dimensional space can be divided in higher dimensional space internal linear.RBF network structure is simple, training is succinct and study fast convergence rate, can Approximation of Arbitrary Nonlinear Function.Therefore RBF network has application comparatively widely: as time series analysis, pattern-recognition, the field such as nonlinear Control and image processing.φ
2) genetic algorithm (Genetic Algorithm, GA) is that evolution laws (survival of the fittest, the survival of the fittest genetic mechanism) that a class uses for reference living nature develops and the randomization searching method that comes.The probability search method of the evolutionary mechanism of its simulation " survival of the fittest, the survival of the fittest ".This algorithm take fitness function as foundation, completing structural rearrangement iteration, can realize a kind of algorithmic procedure of adaptive global optimization probabilistic search by carrying out genetic manipulation to population at individual.Compared with other optimized algorithms, there is computing time few, the advantages such as the high and fast convergence rate of robustness, therefore, genetic algorithm it be widely used in the fields such as function optimization, automatically control, parameter optimization, production management, artificial neural network, optimum control.
The present embodiment adopts GA to search for the connection weights of radial primary function network, function center and width, asking for can the network structure that maps of adaptive input and output and parameter, simultaneously under the network parameter trained, according to optimization aim, the adjustable input asked for further under this optimal objective is separated.
In engineering reality, from the view point of global optimization, consider comprehensively and coordinate the relation between multiple target, by mathematics and optimized calculation method, under the prerequisite that by-end meets security and environmental requirement, the coal consumption pursued as main target reaches optimum, is shown below:
Wherein, η is boiler net coal consumption rate; NOx
maxfor the upper limit of NOx displacement control.
3) after MATHEMATICAL MODEL OF COMBUSTION is set up, searching process comprehensively adopts genetic algorithm and self adaptation multi-dimensional search method to realize.This problem definition is: under the radial primary function network parameter that existing training completes, and finds the adjustable input combination of suitable boiler combustion system and optimal solution, makes combustion process coal consumption minimum, meet the requirement of the NOx emission upper limit simultaneously.Genetic algorithm can ask for the globally optimal solution under Prescribed Properties, ensures that adjustable input solution all meets the restriction of its minimax, ensures boiler operatiopn security.
Genetic algorithm is the phenomenon such as breeding, mating, variation, evolution in simulating nature circle biological heredity process, take fitness function as foundation, by selecting repeatedly population at individual, copying, intersect, the iterative process such as variation, and finally search the optimal solution of evaluation function.After aforesaid operations, obtain follow-on solution, and progressively eliminate the low solution of fitness function value, increase the solution that fitness function value is high, will evolve after N generation like this optimizes the maximum solution of fitness function value.Shown in the FB(flow block) 4 that figure below is genetic algorithm:
Least mean-square error (the Minimum Square Error of the present embodiment structure actual coal consumption of boiler and radial primary function network outlet chamber, MSE) be cost function, then utilize genetic algorithm to train the center of RBF, width and to be connected weights, enable the input and output data of the adaptive boiler combustion system of the networks function after training, that is, the process of combustion system modeling is realized.Next, on the basis of the radial primary function network after having trained (model parameter is constant), according to the coal consumption target proposed, under the constraint of NOx discharge capacity and each adjustable input quantity range constraint, further employing genetic algorithm searches for suitable adjustable input quantity, for the burning optimization operation at scene provides guidance.
The following describes self adaptation multidimensional search algorithm.
Adopt above-mentioned genetic algorithm for solving, if there is the situation without solution or multiple solution, by the input solution in self adaptation multidimensional search algorithm substep solving-optimizing direction, ensure that the combustion process under regulating by a small margin is stablized.
The benefit of self adaptation multi-dimensional search method: 1, when defining search time, self adaptation multi-dimensional search method has less amount of calculation than other algorithms, can produce reasonable result, therefore has reasonable real-time.Especially real-time online optimization during boiler operatiopn is applicable to.2, reversibility is relatively good: time disconnected in the algorithm, can return a feasible solution, can the security that runs of safeguards system.Search enters in dangerous operation area or NOx discharge capacity and prescribes a time limit, and algorithm can shift to an earlier date and automatically stop, and like this, achieves the boiler efficiency optimization under multi-constraint condition.3, can random algorithm be introduced, next node of random selecting, at this node, detect the number of times that object function improves, then determine whether really to move to this node.
Multi-dimensional search method adopts iterative computation, first select randomly a feasible solution as current solution p (this solution is also referred to as the starting point of multi-dimensional search method), and then in field, select another feasible solution p ', if meet fitness function fitness (p ') > fitness (p) (can be boiler efficiency or NOx concentration), then with p ' replacement p as current solution, otherwise in field, select new p ' else make comparisons with current solution.According to the continuous iteration of above-mentioned rule, until can not the solution that fitness is larger be found in the field of current solution, by current solution p
bestexport as optimal solution.The specific strategy choosing the next feasible solution p ' of multi-dimensional search method is:
Tieing up solution space at N, there is 2N the mutually orthogonal direction of search in current solution p, and the hypersphere centered by p is evenly chosen 2N some formation grid, the expression formula of mesh node is
p
nodes＝p+r·column{[I
N×N-I
N×N]}
R is hypersphere radius, and column{} represents and gets matrix column.Calculate the fitness of all nodes
and by it compared with fitness fitness (p) of current solution, the result compared can be divided into two kinds of situations:
Situation 1: have the fitness of L node to be greater than the fitness of current solution p '
if extreme value is regarded as on one " mountain peak ", this situation means that current solution is positioned on " hillside ", according to parallelogram law, the gradient direction of p can be equivalent to grads=α/| α |, wherein,
The renewal p ' then separated=p+rgrad, r remains unchanged, and then re-establishes grid, proceeds to next step iteration.
Situation 2: the fitness of all nodes is all less than or equal to the fitness of current solution, i.e. L=0, this means that current solution is positioned near certain mountain peak, and the distance on current solution and mountain peak is not more than certain threshold value set.If r > is r
min, make r=α × r, p is constant, re-establishes grid; Otherwise end program, and using current p as optimal solution p
bestexport.Here, α is mesh radius shrinkage factor, r
minit is the search precision of setting.
For high dimensional nonlinear function, the result of its gradient calculation is often very complicated, the fitness of multi-dimensional search method mesh node replaces gradient calculation, reduce operand, and compared with common climbing method, multi-dimensional search method ensure that each iteration current solution p is always with the fastest direction approximation extreme point, higher with this search efficiency.
Below, illustrate how the present embodiment pulverized coal power boiler combustibility method for on-line optimization combines with existing pulverized coal power boiler control system.
As shown in Figure 5, inner ring is the original DCS closed-loop control system of boiler of power plant to boiler performance principle of optimality block diagram, and the initial data that its operating parameter setting value is provided through thermal test by designing unit is determined.But along with the change (as ature of coal, fineness of pulverized coal, coal blending mode, air distribution mode, boiler load change, equipment rebuilding etc.) of boiler operatiopn inside and outside portion environment, this group set-point of boiler control system is often difficult to the system model after Adaptive change, the situation that system cannot be run by design objective is very general, and direct result is that boiler efficiency declines, pollutant emission increases.
Therefore, the closed loop security performance only ensureing boiler operatiopn is inadequate, still need configure boiler performance dynamic optimization system at its exterior, this system is to be accurately recognized as basis to boiler Mathematical Modeling, utilize the mathe-matical map relation of its adjustable parameter (input) and intermediate parameters and reflection parameter (quantity of state), dynamic conditioning boiler DCS operation with closed ring parameter set-point, under the prerequisite that security is allowed, the open loop of combustion system is even regulated to input, the degree of depth excavates service ability and the running space of steam generator system, under the prerequisite that NOx discharge does not exceed standard, realize the optimized running of boiler combustion efficiency.
Boiler combustion process is often subject to the impact of the working conditions change such as ature of coal, environment temperature, boiler plant overhaul.Under these perturbation actions, the input of boiler combustion model, the quantitative relation of output also change thereupon.If former optimization system design can not adapt to the change of this operating mode, then there is the risk of inefficacy.That is, accessible Optimal State under former operating mode, under new operating mode, regression is suboptimum or unexcellent running status.Therefore, very necessaryly kainogenesis data are rationally incorporated in the historical data, the original combustion system model of regular update, to adapt to contingent working conditions change at any time.The renewal process of model depends on the system of selection of training inputoutput data, has contained the input/output relation under history boiler operatiopn external condition, under historical conditions, can be used for the optimizing process of history in history boiler operatiopn data.But, when boiler exterior operating mode or internal structure parameter change, the data of current or nearest generation must be incorporated to upgrade training data set.According to the situation that the time from the close-by examples to those far off changes, give the forgetting factor that data are certain, it is higher that the data far away apart from current time forget degree, and current data then " without forgeing " incorporates training dataset.Like this, can ensure that the network after online updating can outside operating mode after " in time " Adaptive change, then, adopt genetic algorithm and self adaptation multidimensional search algorithm to realize dynamic optimization.
Boiler performance optimization does last layer Supervised Control on the basis of original DSC control system, only the operation of DCS is advised, do not change the existing hardware system of boiler, therefore the safety and stability that boiler of power plant runs can be ensured, as shown in Figure 6, boiler performance optimization has the function of comprehensive coordinate, directly can provide optimizing operation scheme or participate in running optimizatin directly to control, and compensate for the deficiency of DCS system.Fig. 7 shows the principle schematic of boiler performance optimization from another angle.
With reference to the global design flow chart that Fig. 8, Fig. 8 are boiler performance optimization.
Boiler operating parameter and present situation are investigated and analysis, formulates the design of concrete Optimum Experiment and debugging plan according to investigation and analysis result, determine optimization aim, optimization range and boundary parameter; Arrange experimental design according to drafted scheme, determine performance optimization basic data acquisition inventory; Work group member according to experimental design, by the basic data under DCS system and pertinent instruments instrument to collect different load operating mode; After basic data acquisition work completes, statistical regression analysis is utilized to analyze test data and screen, the high-quality test data obtained after screening is used to guide software engineer and carries out radial primary function network training, finally determine steam generator system optimized mathematical model, and online flying marking amount measurement data and NOx online monitoring data are introduced in boiler performance optimization system, as on-line optimization foundation as feedback signal.Optimizing operation result is through prediction and checking, under instructional model is optimized in open loop, to the prompting of boiler operatiopn personnel real-time release optimizing operation, and boiler operatiopn is provided to optimize policy paper and optimizing operation curve (different load working condition) to professional and technical personnel; Under closed optimized control pattern, optimize instruction and directly export optimizing operation instruction by DCS.
In one more specifically embodiment, according to following steps, coal powder boiler combustion process is optimized:
1, analyze the optimizing redundancy degree of existing Combustion System of Boiler Burning Fine, under the guidance of optimum the principle of experimental design, design orthogonal experiment, completes Optimal Experimental in early stage.
2, utilize multivariable nonlinearity correlation analysis to carry out data screening, select the input data that same optimization aim amount causal correlation degree is high, determine input variable set that is rational, simplifying.The multiple filtering algorithm of comprehensive employing carries out pretreatment to inputoutput data, to remove noise contribution wherein and Outliers in Data Processing.
3, according to the experimental data after process, radial primary function network is utilized to carry out modeling to boiler input, outlet chamber mathe-matical map, adopt genetic algorithm training and optimized network parameter simultaneously, make it can mate and reflect Nonlinear Mapping relation between the input and output of existing Combustion System of Boiler Burning Fine.
This method has following advantage:
The first, be difficult to measure and the caused offline optimization result Problem of Failure that often changes for ature of coal, propose the Rolling optimal strategy combined with global optimizing and local search, adopt genetic algorithm and self adaptation multi-dimensional search method to realize on-line optimization; Use statistical method to consider historical data and the contribution newly ceased training data, rationally forget the contribution of partial history data to network training, to adapt to the impact of coal varitation on off-line training optimization system, improve its effect.
The second, can carry out the optimization of steam generator system all round properties according to user's request, optimizing algorithm self has the ability solving global optimum's running status under constraints.According to different emphasis, improve boiler efficiency, reduce net coal consumption rate, prevention or improvement coking and slagging, reduction NOx emission, improve the security of boiler operatiopn, reliability and economy.The enforcement of these Boiler Combustion Optimization System technical measures is to power plant's existing equipment system without any destruction, and the system improved and original system can switchover operations, do not have adverse effect to Power Plant safe operation.
In boiler combustion system modeling process, other mathematical method can also be used, as the object using linear regression model (LRM), ARTOICAL NEURAL NETWORK MODEL, supporting vector machine model etc. all can realize modeling.But on the generalization ability of modeling accuracy and network and the present invention's difference to some extent.After obtaining boiler burnup Mathematical Modeling, the Optimization Solution process that model exports target can adopt multiple optimized algorithm optimizing, as: traditional climbing method, gradient method, steepest descent method, ant group algorithm, particle swarm optimization algorithm etc.But efficiency of algorithm, engineering realizability, find the overall situation solve most in ability etc. with the method applied in the present invention difference to some extent.
On the other hand, present invention also offers a kind of preferred embodiment of pulverized coal power boiler combustibility on line optimization system, with reference to shown in Fig. 9.The present embodiment pulverized coal power boiler combustibility on line optimization system comprises: basic data acquisition module 10, Mathematical Modeling determination module 20 and regulated variable input combination and optimal value determination module 30.During concrete enforcement, the variable data recorded in Combustion System of Boiler Burning Fine EDNA database is screened, analyzed by nonlinear correlation, obtain and affect 100-200 higher variable of the degree of correlation of boiler efficiency and NOx, then 3 groups: 1 group is divided into be that input variable (or is called regulated variable, independent in scope adjustment can be limit in safety), comprise the baffle openings such as First air, Secondary Air, burnout degree, surrounding air, racemization wind, burner nozzle angle, the instantaneous coal-supplying amount of feeder, unit load, coal qualities test index etc.2 groups is that state variable (can reflect the intermediate variable that burning is good and bad, affect significantly by input variable, simultaneously can direct explicit decision optimization aim amount), comprise flue gas oxygen content, unburned carbon in flue dust, exhaust gas temperature, main steam pressure, main steam temperature, reheat steam temperature, desuperheating water of superheater amount, reheater spray water flux, pressure fan electric current, air-introduced machine electric current etc.3 groups is output variable (also known as optimization aim amount), comprises net coal consumption rate (g/kWh) and NOx (mg/m
3).Wherein, net coal consumption rate " can affect table about boiler parameter change to net coal consumption rate " by state variable by U.S.'s DianKeYuan and determine weights, and weighted sum obtains.Like this, namely optimization problem may be defined under the environmental requirement of NOx emission concentration limits, seek net coal consumption rate minimum time, the optimum combination of corresponding input variable and optimum value, that is,
Wherein, basic data acquisition module 10 is in the basic data under different load operating mode for the boiler receiving DCS system and instrument and meter collection, write the real-time data base software interfaces such as Power Plant DCS System conventional EDNA, openplant, pi, the various variate-values (comprising all input variables, state variable, output variable) affecting boiler combustion situation are read from database medium time interval, and stored in MySql database to be optimized analysis.
Mathematical Modeling determination module 20 is for foundation basic data, and employing radial primary function network sets up the Nonlinear Mapping relation between the adjustable input variable of pulverized-coal fired boiler and output variable, and Nonlinear Mapping relation is as the Mathematical Modeling of boiler combustion.Networks function be input as regulated variable, output is two optimization aim variablees.The hidden layer activation primitive of RBF neural is a kind of kernel function of radial symmetric.When input amendment propagates into hidden unit space, this group kernel function constitutes one group " base " of input amendment.When input signal is near basic function central range, hidden layer node by producing larger output, so radial primary function network is also referred to as local sensing field network.From function input-output characteristic, only have when in the certain area that input vector falls into RBF neurone clustering immediate vicinity, this neuronic transfer function just can be made to produce larger output, and namely transfer function serves the effect of mode detector.Choosing for kernel function, has multiple choices (e.g., Gaussian function, nucleus vestibularis triangularis function, two exponential kernel functions etc.).This patent adopts gaussian kernel function.The hidden layer of RBF network realizes basic function to the Linear Mapping exported to output layer, and the output namely corresponding to a kth output node is the quadratic approach that each node of hidden layer exports.According to above constructed network structure, this network can be trained to make it mate N dimension input data x
n(n=1,2 ..., N) and export data y to K dimension
k(k=1,2 ..., K) between mapping relations.In typical case, radial basis function network hidden neuron quantity I should be less than or equal to input vector quantity N.During I<N, then need to carry out cluster (self-organized learning method) to Basis Function Center, use K-means clustering algorithm to find I RBF center.The width of basic function, once determine just to secure, then will be determined in center.When RBF selects Gaussian function, bandpass is desirable,
wherein, d
maxfor the ultimate range between selected center.Also find suitable basic function width by optimizing mode, that is, suitable basic function width should make output error and minimum.
Study weights
) study of weights can use least square method (LeastMean Square, LMS) method, LMS algorithm 2 of should be noted that are: the first, LMS algorithm is input as the output of RBF network hidden layer; The output weighted sum of neuron just to hidden neuron of the second, RBF network output layer.The study of weights adopts the LMS method of Gradient Descent iteration.
The application adopts genetic algorithm to try to achieve optimal base function widths and the hidden layer weight matrix to output layer.Genetic algorithm is the phenomenon such as breeding, mating, variation, evolution in simulating nature circle biological heredity process, take fitness function as foundation, by selecting repeatedly population at individual, copying, intersect, the iterative process such as variation, and finally search the optimal solution of evaluation function.Utilize genetic algorithm to optimal base function widths σ
i(i=1,2 ..., I) and hidden layer to the weight matrix w of output layer
ki(k=1,2 ..., K; I=1,2 ..., P) and optimizing step is as follows:
A) optimized variable and constraints is determined: select σ
iparameter area is 0 ~ 1, w
kiparameter area is 0 ~ 1.
B) chromosomal encoding and decoding: use 10 binary strings to represent individuality in colony, produced the gene of the individuality in initial population by equally distributed random number.
C) interpretational criteria of ideal adaptation degree: in view of net coal consumption rate 1/ η and nitrous oxides concentration NOx is the main aspect weighing boiler combustion system performance quality, therefore, this patent selects simple target under a constraints as evaluation function
D) genetic operator is designed: selection algorithm adopts " roulette algorithm " selective staining body to produce the next generation; Crossing operation uses single-point crossover operator the part-structure of two parent individualities replaced restructuring and generate new individual; Mutation operator uses basic bit mutation operator (that is, changing some gene location of the individuality string in colony).
E) determine the operational factor of genetic algorithm: group size is 80, iterations is 100, and crossover probability is 0.9, and mutation probability is 0.05.
After aforesaid operations, obtain follow-on solution, and progressively eliminate the low solution of fitness function value, increase the solution that fitness function value is high, will evolve after N generation like this optimizes the maximum solution of fitness function value.
Regulated variable input combination and optimization solution determination module 30, for according to this Mathematical Modeling, obtain corresponding expectation coal consumption and NO
xthe input combination of boiler combustion system regulated variable and optimal solution under emission level.
Regulated variable input combination and optimization solution determination module 30 comprise the first optimizing unit 301 further, this the first optimizing unit 301 is for adopting genetic algorithm, under radial primary function network parameter, find the adjustable input combination of suitable boiler combustion system and optimal solution, make combustion process coal consumption minimum, meet the requirement of the NOx emission upper limit simultaneously.Utilize the determined input and output mapping model of radial primary function network in Mathematical Modeling determination module 20, keep the basic function width cs trained
iwith hidden layer to output layer weight w
kiconstant.New optimization problem is configured to: seek adjustable input value suitable in safe range, make
further adjustable input variable is carried out binary coding, the genetic algorithmic steps in replicated blocks 20, the adjustable input value of optimum under lower net coal consumption rate level can be obtained.
Regulated variable input combination separate optimal solution determination module comprise further the second optimizing unit 302: the second optimizing unit 302 for genetic algorithm determine the adjustable input of boiler combustion system combine and optimal solution time, if without separating or having multiple solution, then continue through self adaptation multidimensional search algorithm and determine the concrete value of each variable in input combination.By Partial Variable composition search volume larger on output variable impact for input variable.Adopt introduced self adaptation multidimensional search algorithm above, quick off-line or find online, make the direction of search that target capabilities is improved, explore the feasible solution existed under the condition of new target capabilities.
Further preferably, Mathematical Modeling determination module 30 is further used for, and the historical data of boiler combustion is regularly incorporated kainogenesis data, and regular update, in order to train radial primary function network, obtains the Mathematical Modeling optimized.The Mathematical Modeling of boiler combustion can change along with the change of external condition (e.g., ature of coal, environment temperature, equipment rebuilding etc.).For adapting to this change, the Mathematical Modeling of its combustion system should be able to constantly online updating.The mode historical data of boiler combustion regularly being incorporated kainogenesis data is from the close-by examples to those far off change according to the time, gives the data forgetting factor of specifying, the data that distance current time is far away to forget degree higher, it is 0 that current data then forgets degree.According to the inverse of forgetting factor, determine the probability size introducing training dataset, nearer data are higher by the possibility as training data, and the data of former generation are lower by the possibility as training data.Off-line training the frequency upgrading combustion system Mathematical Modeling will determine according to the ruuning situation of actual set.
Because the principle of pulverized coal power boiler combustibility on line optimization system is similar to pulverized coal power boiler combustibility method for on-line optimization, sufficient explanation is done to pulverized coal power boiler combustibility method for on-line optimization above, therefore, do not repeat them here, relevant part illustrates with reference to aforementioned.
Obviously, those skilled in the art can carry out various change and modification to the present invention and not depart from the spirit and scope of the present invention.Like this, if these amendments of the present invention and modification belong within the scope of the claims in the present invention and equivalent technologies thereof, then the present invention is also intended to comprise these change and modification.
Claims (10)
1. a pulverized coal power boiler combustibility method for on-line optimization, is characterized in that, comprises the steps:
Step 1, the boiler receiving DCS system and instrument and meter collection is in the basic data under different load operating mode;
Step 2, according to described basic data, employing radial primary function network sets up the Nonlinear Mapping relation between the adjustable input variable of pulverized-coal fired boiler and output variable, and described Nonlinear Mapping relation is as the Mathematical Modeling of boiler combustion;
Step 3, according to this Mathematical Modeling, obtains corresponding expectation coal consumption and NO
xthe input combination of boiler combustion system regulated variable and optimal value under emission level.
2. pulverized coal power boiler combustibility method for on-line optimization according to claim 1, it is characterized in that, described step 3 comprises further:
Adopt genetic algorithm, under described radial primary function network parameter, find the adjustable input combination of suitable boiler combustion system and optimal solution, make combustion process coal consumption minimum, meet the requirement of the NOx emission upper limit simultaneously.
3. pulverized coal power boiler combustibility method for on-line optimization according to claim 2, is characterized in that,
Described genetic algorithm, when determining the adjustable input combination of boiler combustion system and optimal solution, if without separating or having multiple solution, then continues through the determination that self adaptation multidimensional search algorithm carries out inputting each variable-value in combination.
4. pulverized coal power boiler combustibility method for on-line optimization according to claim 3, is characterized in that,
Described step 3 comprises further, by boiler combustion historical data regularly incorporate kainogenesis data, regular update in order to train described radial primary function network, obtain optimize Mathematical Modeling.
5. pulverized coal power boiler combustibility method for on-line optimization according to claim 4, is characterized in that, described by boiler combustion historical data regularly incorporate kainogenesis data and be further,
From the close-by examples to those far off change according to the time, give the data forgetting factor of specifying, the data that distance current time is far away to forget degree higher, it is 0 that current data then forgets degree.
6. a pulverized coal power boiler combustibility on line optimization system, is characterized in that, comprising:
Basic data acquisition module, the boiler for receiving DCS system and instrument and meter collection is in the basic data under different load operating mode;
Mathematical Modeling determination module, for according to described basic data, adopt radial primary function network to set up Nonlinear Mapping relation between the adjustable input variable of pulverized-coal fired boiler and output variable, described Nonlinear Mapping relation is as the Mathematical Modeling of boiler combustion;
Regulated variable input combination and optimal value determination module, for according to this Mathematical Modeling, obtain corresponding expectation coal consumption and NO
xthe input combination of boiler combustion system regulated variable under emission level.
7. pulverized coal power boiler combustibility on line optimization system according to claim 6, is characterized in that, described regulated variable input combination determination module comprises further:
First optimizing unit, for adopting genetic algorithm, under described radial primary function network parameter, finding the adjustable input combination of suitable boiler combustion system and optimal solution, making combustion process coal consumption minimum, meet the requirement of the NOx emission upper limit simultaneously.
8. pulverized coal power boiler combustibility on line optimization system according to claim 7, is characterized in that, described regulated variable input combination determination module comprises further:
Second optimizing unit, described genetic algorithm, when determining the adjustable input combination of boiler combustion system and optimal solution, if without separating or having multiple solution, then continues through self adaptation multidimensional search algorithm and determines each variable-value in input combination.
9. pulverized coal power boiler combustibility on line optimization system according to claim 8, is characterized in that,
Described Mathematical Modeling determination module is further used for, and the historical data of boiler combustion is regularly incorporated kainogenesis data, and regular update, in order to train described radial primary function network, obtains the Mathematical Modeling upgraded.
10. pulverized coal power boiler combustibility on line optimization system according to claim 9, is characterized in that, described by boiler combustion the historical data mode that regularly incorporates kainogenesis data be,
From the close-by examples to those far off change according to the time, give the data forgetting factor of specifying, the data that distance current time is far away to forget degree higher, it is 0 that current data then forgets degree.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510094539.2A CN104763999A (en) | 2015-03-04 | 2015-03-04 | Power plant pulverized coal boiler combustion performance online optimizing method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510094539.2A CN104763999A (en) | 2015-03-04 | 2015-03-04 | Power plant pulverized coal boiler combustion performance online optimizing method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104763999A true CN104763999A (en) | 2015-07-08 |
Family
ID=53646002
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510094539.2A Pending CN104763999A (en) | 2015-03-04 | 2015-03-04 | Power plant pulverized coal boiler combustion performance online optimizing method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104763999A (en) |
Cited By (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106168381A (en) * | 2016-07-07 | 2016-11-30 | 神华集团有限责任公司 | The control system of boiler combustion, method and apparatus |
| CN108446459A (en) * | 2018-03-01 | 2018-08-24 | 云南师范大学 | Process of coking heat consumption influence factor optimization method based on fuzzy semantics reasoning |
| CN109063359A (en) * | 2018-08-16 | 2018-12-21 | 燕山大学 | A kind of dynamic modelling method of Process of Circulating Fluidized Bed Boiler |
| CN109828461A (en) * | 2019-01-24 | 2019-05-31 | 东南大学 | A kind of thermal technology's multiloop controller iterative adjusting system and setting method based on population |
| CN110268349A (en) * | 2017-02-10 | 2019-09-20 | 三菱日立电力系统株式会社 | Test device for planning and test planing method |
| CN110501983A (en) * | 2019-07-31 | 2019-11-26 | 农业农村部南京农业机械化研究所 | Expert control system and control method based on batch seed-coating machine |
| CN110568766A (en) * | 2019-10-22 | 2019-12-13 | 华北电力大学 | Boiler control system based on combustion layer temperature and online optimization method |
| CN110703598A (en) * | 2019-09-16 | 2020-01-17 | 中国神华能源股份有限公司国华电力分公司 | Unit operation parameter optimizing method and device, electronic equipment and storage medium |
| CN110824099A (en) * | 2019-11-07 | 2020-02-21 | 东南大学 | Method for predicting reaction performance in solid fuel chemical chain process based on GBRT |
| CN110837226A (en) * | 2019-12-26 | 2020-02-25 | 华润电力技术研究院有限公司 | Thermal power generating unit operation optimization method based on intelligent optimization algorithm and related device |
| CN110941186A (en) * | 2019-12-26 | 2020-03-31 | 华润电力技术研究院有限公司 | Steam temperature control optimization method based on neural network and universal gravitation search algorithm |
| CN110986085A (en) * | 2019-11-21 | 2020-04-10 | 东南大学 | Four-corner tangential boiler combustion optimization control method based on air distribution mode optimization |
| CN110981240A (en) * | 2019-12-19 | 2020-04-10 | 华东理工大学 | Calcination process optimization method and system |
| CN111199304A (en) * | 2018-11-19 | 2020-05-26 | 天津市职业大学 | Multi-target combustion optimization method based on data-driven fusion strategy |
| CN111260107A (en) * | 2018-11-30 | 2020-06-09 | 斗山重工业建设有限公司 | Boiler combustion optimization system and method |
| CN111612211A (en) * | 2020-04-10 | 2020-09-01 | 杭州电子科技大学 | Predictive modeling method for coal powder fineness of coal mill |
| CN111765445A (en) * | 2020-07-01 | 2020-10-13 | 河北工业大学 | Boiler on-line combustion optimization control method and system and computer equipment |
| CN111859774A (en) * | 2019-04-29 | 2020-10-30 | 新奥数能科技有限公司 | Method and device for regulating and controlling air supply system of gas boiler |
| CN112066411A (en) * | 2020-09-02 | 2020-12-11 | 常州工学院 | Optimization method for boiler combustion |
| CN112381289A (en) * | 2020-11-13 | 2021-02-19 | 重庆电力高等专科学校 | Method for improving characteristics of nitrogen oxides based on coal blending combustion boiler |
| CN112418664A (en) * | 2020-11-20 | 2021-02-26 | 西安热工研究院有限公司 | Particle swarm algorithm-based bin-divided combined mixed combustion method and system |
| CN115595390A (en) * | 2022-10-27 | 2023-01-13 | 中冶赛迪信息技术（重庆）有限公司（Cn） | Method and device for calculating coal blending scheme of blast furnace coal injection and electronic equipment |
| WO2024060488A1 (en) * | 2022-09-21 | 2024-03-28 | 浙江大学 | Method based on deep recurrent neural network and evolutionary computation for optimizing combustion of industrial boiler |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR100230855B1 (en) * | 1996-12-03 | 1999-11-15 | 이해규 | Control system configuration method of a small boiler |
| CN2641472Y (en) * | 2003-05-16 | 2004-09-15 | 浙江大学 | On-line optimization control device for boiler combustion based on computation intelligence |
| CN101158988A (en) * | 2007-11-15 | 2008-04-09 | 浙江大学 | Astringe detecting probe method used for large scale varying duty real time process flow simulate |
| CN101799848A (en) * | 2010-03-09 | 2010-08-11 | 江西省电力科学研究院 | Method for obtaining energy loss analysis parameter answer value of furnace of thermal power set |
| CN102226904A (en) * | 2011-05-25 | 2011-10-26 | 浙江省电力试验研究院 | Soft measurement method for air quantity of primary air of large-scale boiler in power station |
| CN103839110A (en) * | 2014-02-24 | 2014-06-04 | 国家电网公司 | Modeling method of prediction of production of nitrogen oxide in boiler |
| CN104331736A (en) * | 2014-11-13 | 2015-02-04 | 国家电网公司 | RBF (Radial Basis Function) neural network based supercritical boiler nitric oxide discharging dynamic predication method |
-
2015
- 2015-03-04 CN CN201510094539.2A patent/CN104763999A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR100230855B1 (en) * | 1996-12-03 | 1999-11-15 | 이해규 | Control system configuration method of a small boiler |
| CN2641472Y (en) * | 2003-05-16 | 2004-09-15 | 浙江大学 | On-line optimization control device for boiler combustion based on computation intelligence |
| CN101158988A (en) * | 2007-11-15 | 2008-04-09 | 浙江大学 | Astringe detecting probe method used for large scale varying duty real time process flow simulate |
| CN101799848A (en) * | 2010-03-09 | 2010-08-11 | 江西省电力科学研究院 | Method for obtaining energy loss analysis parameter answer value of furnace of thermal power set |
| CN102226904A (en) * | 2011-05-25 | 2011-10-26 | 浙江省电力试验研究院 | Soft measurement method for air quantity of primary air of large-scale boiler in power station |
| CN103839110A (en) * | 2014-02-24 | 2014-06-04 | 国家电网公司 | Modeling method of prediction of production of nitrogen oxide in boiler |
| CN104331736A (en) * | 2014-11-13 | 2015-02-04 | 国家电网公司 | RBF (Radial Basis Function) neural network based supercritical boiler nitric oxide discharging dynamic predication method |
Cited By (33)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106168381A (en) * | 2016-07-07 | 2016-11-30 | 神华集团有限责任公司 | The control system of boiler combustion, method and apparatus |
| CN110268349A (en) * | 2017-02-10 | 2019-09-20 | 三菱日立电力系统株式会社 | Test device for planning and test planing method |
| CN108446459A (en) * | 2018-03-01 | 2018-08-24 | 云南师范大学 | Process of coking heat consumption influence factor optimization method based on fuzzy semantics reasoning |
| CN108446459B (en) * | 2018-03-01 | 2022-03-22 | 云南师范大学 | Fuzzy semantic reasoning-based coking process heat consumption influence factor optimization method |
| CN109063359A (en) * | 2018-08-16 | 2018-12-21 | 燕山大学 | A kind of dynamic modelling method of Process of Circulating Fluidized Bed Boiler |
| CN111199304B (en) * | 2018-11-19 | 2023-11-14 | 天津市职业大学 | Multi-target combustion optimization method based on data-driven fusion strategy |
| CN111199304A (en) * | 2018-11-19 | 2020-05-26 | 天津市职业大学 | Multi-target combustion optimization method based on data-driven fusion strategy |
| CN111260107A (en) * | 2018-11-30 | 2020-06-09 | 斗山重工业建设有限公司 | Boiler combustion optimization system and method |
| CN111260107B (en) * | 2018-11-30 | 2024-03-05 | 斗山重工业建设有限公司 | Boiler combustion optimization system and method |
| CN109828461B (en) * | 2019-01-24 | 2021-09-07 | 东南大学 | Closed-loop iteration particle swarm setting system and method for thermotechnical multi-loop controller |
| CN109828461A (en) * | 2019-01-24 | 2019-05-31 | 东南大学 | A kind of thermal technology's multiloop controller iterative adjusting system and setting method based on population |
| CN111859774B (en) * | 2019-04-29 | 2024-05-14 | 新奥数能科技有限公司 | Method and device for regulating and controlling air supply system of gas boiler |
| CN111859774A (en) * | 2019-04-29 | 2020-10-30 | 新奥数能科技有限公司 | Method and device for regulating and controlling air supply system of gas boiler |
| CN110501983B (en) * | 2019-07-31 | 2021-03-26 | 农业农村部南京农业机械化研究所 | Expert control system and control method based on batch-type coating machine |
| CN110501983A (en) * | 2019-07-31 | 2019-11-26 | 农业农村部南京农业机械化研究所 | Expert control system and control method based on batch seed-coating machine |
| CN110703598A (en) * | 2019-09-16 | 2020-01-17 | 中国神华能源股份有限公司国华电力分公司 | Unit operation parameter optimizing method and device, electronic equipment and storage medium |
| CN110568766A (en) * | 2019-10-22 | 2019-12-13 | 华北电力大学 | Boiler control system based on combustion layer temperature and online optimization method |
| CN110824099A (en) * | 2019-11-07 | 2020-02-21 | 东南大学 | Method for predicting reaction performance in solid fuel chemical chain process based on GBRT |
| CN110824099B (en) * | 2019-11-07 | 2022-03-04 | 东南大学 | Method for predicting reaction performance in solid fuel chemical chain process based on GBRT |
| CN110986085A (en) * | 2019-11-21 | 2020-04-10 | 东南大学 | Four-corner tangential boiler combustion optimization control method based on air distribution mode optimization |
| CN110986085B (en) * | 2019-11-21 | 2021-07-27 | 东南大学 | Four-corner tangential boiler combustion optimization control method based on air distribution mode optimization |
| CN110981240A (en) * | 2019-12-19 | 2020-04-10 | 华东理工大学 | Calcination process optimization method and system |
| CN110941186A (en) * | 2019-12-26 | 2020-03-31 | 华润电力技术研究院有限公司 | Steam temperature control optimization method based on neural network and universal gravitation search algorithm |
| CN110837226A (en) * | 2019-12-26 | 2020-02-25 | 华润电力技术研究院有限公司 | Thermal power generating unit operation optimization method based on intelligent optimization algorithm and related device |
| CN111612211A (en) * | 2020-04-10 | 2020-09-01 | 杭州电子科技大学 | Predictive modeling method for coal powder fineness of coal mill |
| CN111765445A (en) * | 2020-07-01 | 2020-10-13 | 河北工业大学 | Boiler on-line combustion optimization control method and system and computer equipment |
| CN111765445B (en) * | 2020-07-01 | 2021-10-15 | 河北工业大学 | Boiler on-line combustion optimization control method and system and computer equipment |
| CN112066411A (en) * | 2020-09-02 | 2020-12-11 | 常州工学院 | Optimization method for boiler combustion |
| CN112381289A (en) * | 2020-11-13 | 2021-02-19 | 重庆电力高等专科学校 | Method for improving characteristics of nitrogen oxides based on coal blending combustion boiler |
| CN112418664B (en) * | 2020-11-20 | 2023-06-20 | 西安热工研究院有限公司 | Particle swarm optimization-based binning combination blending method and system |
| CN112418664A (en) * | 2020-11-20 | 2021-02-26 | 西安热工研究院有限公司 | Particle swarm algorithm-based bin-divided combined mixed combustion method and system |
| WO2024060488A1 (en) * | 2022-09-21 | 2024-03-28 | 浙江大学 | Method based on deep recurrent neural network and evolutionary computation for optimizing combustion of industrial boiler |
| CN115595390A (en) * | 2022-10-27 | 2023-01-13 | 中冶赛迪信息技术（重庆）有限公司（Cn） | Method and device for calculating coal blending scheme of blast furnace coal injection and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104763999A (en) | Power plant pulverized coal boiler combustion performance online optimizing method and system | |
| CN105020705A (en) | Method and system for optimizing and controlling combustion performance of circulating fluidized bed boiler in real time | |
| Wang et al. | Prediction of NOx emission for coal-fired boilers based on deep belief network | |
| Rahat et al. | Data-driven multi-objective optimisation of coal-fired boiler combustion systems | |
| CN106931453B (en) | The forecasting system and method for circulating fluid bed domestic garbage burning emission of NOx of boiler | |
| US11232376B2 (en) | System and method for optimizing combustion of boiler | |
| Zhan et al. | Deepthermal: Combustion optimization for thermal power generating units using offline reinforcement learning | |
| CN106224939B (en) | Circulating fluid bed domestic garbage burning boiler bed temperature Forecasting Methodology and system | |
| CN107038334A (en) | Circulating fluid bed domestic garbage burning boiler CO emitted smoke system and methods | |
| CN106055520B (en) | System and method for predicting drum water level of circulating fluidized bed domestic garbage incineration boiler | |
| CN107726358A (en) | Boiler Combustion Optimization System and method based on CFD numerical simulations and intelligent modeling | |
| Hou et al. | Coordinated control system modeling of ultra-supercritical unit based on a new fuzzy neural network | |
| CN103728879A (en) | Power station boiler emission soft measuring method based on least squares support vector machine and on-line updating | |
| CN107016455A (en) | The forecasting system and method for circulating fluid bed domestic garbage burning boiler furnace outlet flue gas oxygen content | |
| CN109670629A (en) | Coal-burning boiler thermal efficiency forecast method based on shot and long term Memory Neural Networks | |
| CN105864797A (en) | Real-time prediction system and method for boiler entering heat value of circulating fluidized bed household garbage incineration boiler | |
| CN103729569A (en) | Soft measurement system for flue gas of power-station boiler on basis of LSSVM (Least Squares Support Vector Machine) and online updating | |
| Cass et al. | Adaptive process optimization using functional-link networks and evolutionary optimization | |
| CN102842066A (en) | Modeling method for combustion optimization of biomass furnace | |
| Savargave et al. | Modeling and optimizing boiler design using neural network and firefly algorithm | |
| Ma et al. | Three-objective optimization of boiler combustion process based on multi-objective teaching–learning based optimization algorithm and ameliorated extreme learning machine | |
| Zhang et al. | A novel operation cost optimization system for mix-burning coal slime circulating fluidized bed boiler unit | |
| Wen et al. | NOx emission predicting for coal-fired boilers based on ensemble learning methods and optimized base learners | |
| Xu et al. | A new online optimization method for boiler combustion system based on the data-driven technique and the case-based reasoning principle | |
| Movahed et al. | Modeling and optimization of NO emission for a steam power plant by data‐driven methods |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| EXSB | Decision made by sipo to initiate substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20150708 |
|
| RJ01 | Rejection of invention patent application after publication |