CN107016455A - The forecasting system and method for circulating fluid bed domestic garbage burning boiler furnace outlet flue gas oxygen content - Google Patents
The forecasting system and method for circulating fluid bed domestic garbage burning boiler furnace outlet flue gas oxygen content Download PDFInfo
- Publication number
- CN107016455A CN107016455A CN201710108399.9A CN201710108399A CN107016455A CN 107016455 A CN107016455 A CN 107016455A CN 201710108399 A CN201710108399 A CN 201710108399A CN 107016455 A CN107016455 A CN 107016455A
- Authority
- CN
- China
- Prior art keywords
- particle
- value
- algorithm
- individual
- population
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
Abstract
The invention discloses a kind of real-time estimate system and method for circulating fluid bed domestic garbage burning boiler furnace outlet flue gas oxygen content, using the method for algorithm of support vector machine and genetic particle colony optimization algorithm integrated moulding on multiple populations, the system and method for constructing a kind of fast, economical and adaptive updates carry out real-time estimate to boiler furnace outlet flue gas oxygen content, avoid the modelling by mechanism work of very complicated.The dynamic variation characteristic of flue gas oxygen content is characterized using the nonlinear dynamic characteristic of SVM algorithm, generalization ability and real-time estimate ability;SVM algorithm punishment parameter C and nuclear parameter g are optimized using particle swarm optimization algorithm, the generalization ability of model is improved;Introduce genetic operation operator and migration mechanism on multiple populations, accelerate the convergence rate of particle cluster algorithm, the diversity of particle swarm optimization algorithm solution is improved, particle cluster algorithm optimizing is reduced and calculates the possibility for being absorbed in local optimum, improve the ability of searching optimum and local search ability of algorithm.
Description
Technical field
The present invention relates to energy project field, especially, it is related to a kind of circulating fluid bed domestic garbage burning boiler furnace
Exiting flue gas oxygen content emitted smoke system and method.
Background technology
Waste incineration is due to can well realize the volume reduction, minimizing, innoxious and recycling of technology of garbage disposal, closely
In more than ten years, under the guiding of national related industry policy, domestic waste incineration industry achieves vigorous growth.From last century
The nineties, domestic many scientific research structures are burnt to Municipal Solid Waste in China (Municipal Solid Waste, MSW)
Mechanism has carried out a large amount of further investigations, has grasped the burning spy of mixed collection, moisture height, the domestic waste of complicated component
Property, the inferior fuel such as coal, gangue recirculating fluidized bed (Circulating Fluidized Bed, CFB) is burnt according to China
On the basis of the development Experience of technology, refuse incinerator of circulating fluid bed is have developed, from Zhejiang University's exploitation in 1998
First fluidized bed refuse incinerator puts into operation beginning, shows suitable for domestic high-moisture, calorific value be relatively low and fluctuation
Property the characteristics of very big house refuse carries out large-scale burning disposal.At present, CFB garbage incineration technologies at home many
Individual city has carried out popularization and application, ends for the end of the year 2015, the built platform of garbage burning boiler more than 70 in the country, day processing quantity of refuse
6.9 ten thousand tons, be that the incineration treatment of garbage industry of China is made that important contribution.
Boiler furnace outlet flue gas oxygen content is to weigh one of whether safe boiler, economy, important symbol of environmental protection operation.
Furnace outlet oxygen content is too low, and the burning in burner hearth can not be carried out fully, and substantial amounts of combustible fails in after-flame, rubbish
Harmful substance also can not be destroyed sufficiently, and the clinker ignition loss for inevitably resulting in clinker exceedes《Consumer waste incineration pollution control
Standard processed》(GB18485-2014) defined minimum standard, burner hearth also faces value and ties soft burnt danger in itself;Furnace outlet contains
Oxygen amount is too high, then excessive cold air can take away temperature in heat in burner hearth, burner hearth and be difficult to remain stable, influence the safety of boiler
Stable operation.Generally the flue gas oxygen content of furnace outlet is monitored using zirconium oxide in industrial production, but in burner hearth
Environment is extremely complicated, often makes zirconium oxide measuring point because abrasion can not normal work.Meanwhile, simple hard ware measure system can not
Find out the operation variation characteristic of furnace outlet flue gas oxygen content.Therefore, the furnace outlet flue gas for building an enough accuracy is oxygen-containing
Amount forecasting system and model tool are of great significance.
Researcher both domestic and external is carried out to the furnace outlet flue gas oxygen content Dynamic Characteristic Modeling of CFBB
Research, mainly there is a following several method.It is a kind of be according to CFB boiler kinetics of combustion, hydrodynamics, heat and mass spy
Property, set up after rational simplified hypothesis, mechanism model is set up by way of mathematical description.This method can be anti-
Reflect the variation tendency of flue gas oxygen content, but as it is assumed that deviation between model and true model and be unable to reach enough accurate
Degree;Another method is on the basis of substantial amounts of bench run or field test, to be set up by the method for regression analysis
Empirical model on flue gas oxygen content dynamic characteristic.This method needs to expend substantial amounts of manpower and materials, and time cost is high, together
Shi Wufa guarantee tests cover all operating modes, with certain limitation;The third method is passed using Fluid Mechanics Computation, calculating
The Reduced mechanisms simulation stove combustion process both thermally and chemically reacted, accurately solves the generation situation of flue gas oxygen content, display
Good effect has very big development potentiality.But this method is primarily limited to the letter of fluid mechanic model and chemical reaction
Change the gap of mechanism and actual conditions, it is necessary to high-end allocation of computer and very long calculating time, therefore in this way
Still in the initial development stage.In addition, the feeding system uniformity of CFB garbage burning boilers is poor, enter the calorific value ripple of stove rubbish
Dynamic property is big, complicated components, polygons are strong, is one of the main difficulty faced in flue gas oxygen content prediction modeling process, it will
Ask set up flue gas oxygen content dynamic performance model that there is good adaptive ability, above-mentioned three kinds of modeling methods are in this respect
Still it has been short of.
With the development of electronic technology, computer technology and information technology, Distributed Control System (Distributed
Control System, DCS) running of CFB life burning boilers is widely used in, include temperature, pressure, flow etc.
Process data including parameter is all completed and must preserved, and is that people recognize comprising abundant procedural information in these historical datas
Know and understand one of important channel of production process, be that the application of Intelligent data mining algorithm is ground with very high tap value
Study carefully and using there is provided excellent hardware and software platform.VC (Vapnik- of the SVMs based on Statistical Learning Theory
Chervonenkis the statistics of theoretical and structural risk minimization (structural risk minimization) principle) is tieed up
Learning method, a SVM key character is exactly the black box characteristic between input and output, it will treat modeling be considered as one it is black
How complicated case, the internal mechanism for being indifferent to problem to be solved is, is only concerned the input and output of system.This causes SVM especially to fit
, in this way can be around for the modeling of the furnace outlet flue gas oxygen content dynamic characteristic of CFB Domestic refuse incinerators
The difficult points such as the hysteresis quality carried in flue gas oxygen content change procedure, non-linear and time variation are opened, flue gas oxygen content and each shadow is realized
Complex mapping relation between the factor of sound.Meanwhile, SVM has self-learning capability, can be trained according to fresh sample, adaptive
Model parameter should be adjusted.When coal, dust stratification and equipment performance (such as performance change of garbage feeding system) change,
Flue gas oxygen content dynamic performance model will also change therewith.Cigarette can be adjusted with on-line training using SVM self-learning property
Gas oxygen content dynamic performance model, it is ensured that the precision of model.
But SVM performance has dependence largely to punishment parameter C and nuclear parameter g, if the two parameters are set
That puts is undesirable, will directly affect the performance of SVM models.In order to improve this problem, this patent is introduced into hereditary grain on multiple populations
Swarm optimization, is optimized with it to parameter C and g.Meanwhile, flue gas oxygen content forecasting system has higher computational load,
Therefore, in order to improve system computational efficiency, it is necessary to rationally set system frame structure.
The content of the invention
In view of the above-mentioned deficiencies in the prior art, it is an object of the present invention to provide a kind of circulating fluid bed domestic garbage burning boiler
The forecasting system and method for furnace outlet flue gas oxygen content.The present invention is in analysis CFB Domestic refuse incinerator operation mechanisms
On the basis of, the input variable of flue gas oxygen content forecast model is selected, modeling is trained to sample set using SVMs, and
Optimizing is carried out to the punishment parameter C and nuclear parameter g of SVM models using Genetic Particle Swarm Algorithm on multiple populations.Effect is calculated in order to improve
Rate and computing resource utilization rate, flue gas oxygen content forecasting system is built using heterogeneous computing environment.
The technical solution adopted for the present invention to solve the technical problems is:A kind of circulating fluid bed domestic garbage burning boiler
The real-time estimate system of furnace outlet flue gas oxygen content.The Distributed Control System and production of the system and CFBB
Management system is connected, including data communication interface and host computer, to flue gas oxygen content forecast model in host computer (server)
It is trained and updates, the model trained is sent to Distributed Control System, production management system (visitor by communication interface after allowing
Family end), the host computer includes:
Signal acquisition module.The module is being burned when specifying house refuse for gathering CFB Domestic refuse incinerators
Operating condition state parameter and performance variable, and constitute refuse thermal value forecast model input variable training sample matrix X (m ×
N), m represents number of samples, and n represents the number of variable;
Data preprocessing module.Gross error processing and random crror processing are carried out to X (m × n), to forgo those not
It is the deceptive information for reflecting boiler accidental conditions, the unusual service conditions such as boiler shutdown, banking fire, batcher blocking is excluded,
In order to avoid the harmful effect that the difference of dimension between the parameter of forecast model and the order of magnitude is caused to model performance, training sample
Input variable is mapped to after normalized in [0,1] interval, the training sample X after being standardized*(m×n).In advance
Processing procedure is carried out using following steps:
1.1) criterion is reached according to Rye, rejects the outlier in training sample X (m × n);
1.2) boiler shutdown operating condition is rejected, the aperture of burner hearth feeder and batcher is zero during boiler shutdown, and
Temperature is close to normal temperature in burner hearth;
1.3) burner hearth banking fire operation conditions, primary air fan, overfire air fan air-introduced machine burner hearth feeder during boiler pressuring fire are rejected
Aperture with batcher is zero, but the temperature of burner hearth emulsion zone maintains 350 DEG C~450 DEG C;
1.4) reject batcher and block operating mode, batcher, which is blocked, needs what operations staff was shot by the camera of material inlet
Picture judges that feed situation when batcher is blocked, operations staff can significantly heighten the aperture of batcher, be reflected in fortune
In row data, i.e., the aperture of batcher is more than 35%;
1.5) data normalization processing.According to formula (1) by the interval of data variable mappings to [0 1].
X in formulaJThe vector that J variables are constituted is represented, min () represents minimum value, and max () represents maximum.
Expertise library module.Training sample is constantly updated using the method for rolling time window so that training sample begins
Be maintained at eventually in newest state, rolling time window method refers to since current time, backtracking L (unit second) length when
Between size.
Model modification determination module.The performance of current predictive model is detected, when relative prediction residual exceedes ± 5%, then
Decision model needs to be updated.
Intelligent modeling module.Intelligent modeling module is the core of flue gas oxygen content forecasting system, the module first with
Genetic Particle Swarm Algorithm on multiple populations carries out optimizing to the punishment parameter C and nuclear parameter g of SVM models, then by obtained optimal ginseng
Array is closed and is assigned to SVM models, and is trained based on this.Algorithm steps are as follows:
2.1) initialization algorithm parameter.The maximum optimizing algebraically T of particle cluster algorithm on multiple populationsmax, maximum inertia weight
ωmax, minimum inertia weight ωmin, speed renewal coefficients R1、R2、R3、R4, population quantity pop, the number of particles of single species
Ind, migration rate Pi, crossover probability Pc, mutation probability Pm, hereditary generation gap Pe, genetic manipulation frequency Pg。
2.2) population is initialized.By the way of real coding, the punishment parameter C and nuclear parameter g of SVM models are compiled in order
Each weight threshold is generated as a real number between [0,200] by code at random among a particle.
2.3) individual extreme value and colony's extreme value are initialized.The parameter combination included in each particle is assigned to SVM models, and
Combined training sample carries out learning training, and the forecast model obtained using training calculates flue gas oxygen content predicted valueBy predicted valueWith actual measured value y*It is compared, and the fitness value fitness using error sum of squares MSE as particle, fitness calculating
Formula is as follows:
Initialization extreme value of the fitness value that the calculating of each particle is obtained as the particle in itself, each population it is central
Value minimum MSE is used as colony's extreme value.
2.4) more new particle.According to newest individual extreme value and global extremum, according to (3) formula and (4) formula more new particle
Speed vid(t) with position xid(t):
vid(t+1)=ω vid(t)+c1r1(pid-xid(t))+c2r2(pgd-xid(t)) (3)
xid(t+1)=xid(t)+vid(t+1) (4)
In formula, t is the optimizing algebraically of particle swarm optimization algorithm, r1, r2It is the random number between [0,1], pidRefer to
I-th of particle is sought so far searches optimal location, pgdRefer to the population and search for optimal location so far.Further, it is
Improve basic particle group algorithm and be easily trapped into local extremum and the slow defect of convergence rate, introduced on the basis of PSO algorithms
Dynamic aceleration pulse c1、c2With inertia weight ω:
Wherein, TmaxFor maximum optimizing algebraically, ωmaxFor maximum inertia weight, ωminFor minimum inertia weight, R1、R2、R3、
R4For constant.
2.5) judge whether to need to carry out genetic manipulation.According to genetic manipulation frequency PgCarry out, generally per PgIn generation, is carried out once
Genetic manipulation.Step 2.6 is then performed to carry out genetic manipulation), otherwise perform step 2.7)
2.6) genetic manipulation is performed.Genetic manipulation includes selecting, intersecting, making a variation, staying excellent, specific as follows.Selection:As wheel disc
Gambling equally calculates select probability, and equidistant selection is individual, if N is to need the individual amount of selection, the distance of select finger is
1/N, the random number of the position of first select finger in [0,1/N] interval is determined;Intersect:The method recombinated using arithmetic,
I.e. two individuals produce new individual by linear combination.Assuming that in two individualsWithBetween carry out arithmetic crossover, then press
Formula (8) and (9) obtain two new individualsWith
λ in formula1And λ2For the real number between (0,1), and λ1+λ2=1;Variation:Uniformity is taken to make a variation, it is assumed that s=(v1,
v2,...,vn) it is parent, z=(z1,z2,...,zn) it is the offspring that variation is produced.One is randomly chosen in present parent vector
Component, it is assumed that be k-th, then in its interval of definition [ak,bk] in take a number v' uniformly randomlykInstead of vkTo obtain zk;
Stay excellent:Using the strategy for retaining parent elite individual, according to the target function value of parent chromosome, replaced with transition for chromosome
P after target function value ranking in parente× 100% chromosome, PeFor replacement rate (hereditary generation gap), in this way, part is most in parent
Excellent individual is retained, and will not be disappeared because of computings such as selection, intersection, variations.
2.7) particle fitness value calculation.The fitness value of particle after updating is calculated according to formula (2).
2.8) more new individual extreme value and colony's extreme value.Using fitness value as evaluation index, relatively more contemporary particle and previous generation
Fitness value size between particle, if the fitness value of current particle is better than previous generation, the position of current particle is set
Individual extreme value is set to, otherwise individual extreme value keeps constant.While the optimal particle of contemporary all particle fitness values is obtained, and with
Previous generation optimal particles are compared, if the fitness value of contemporary optimal particle is better than the fitness of previous generation optimal particles
Value, then be set to global optimum by the adaptive optimal control angle value of contemporary particle, and otherwise global optimum keeps constant.
2.9) immigrant's operation.In nature, the different populations of a species is distributed in different regions, on the one hand each
Groupy phase has facilitated their different existence to independently striving for resource from nature used in oneself under different regional conditions
Pattern and evolution degree;On the other hand, possibly through migrating between each population, make to contact each other, reach mutually
Be connected with without, learn from other's strong points to offset one's weaknesses, the purpose of common evolutionary.(Multi-Population, MP) on multiple populations hereditary grain that the present invention is used
Swarm optimization has exactly used for reference this phenomenon of generally existing in nature.Each time in searching process, using single between population
The excellent individual in immigrant's operation, the 1st population is carried out to the mode of circulation migration and moves to the 2nd, and the 2nd is moved to the 3rd, with
This analogizes, and to the last one is moved to first.Migration rate P between populationiP before ranking in=0.04, i.e. expression source populationi
P after ranking among × 100% individual replacement target populationi× 100% individual, optimal knowledge between population is completed with this
Exchange.
2.10) algorithm stop condition algorithm judges.Judge whether to reach maximum iteration or reach precision of prediction
It is required that, the return to step 2.4 if being not reaching to), continued search for using the particle of renewal, otherwise exit search, perform step
2.11)。
2.11) optimal parameter combination particle is exported.
2.12) parameter combination among optimal particle is assigned to SVM models, and combined training sample is learnt.
2.13) precision of prediction of model is verified.The predicted value and actual value of model are contrasted, relative prediction is calculated and misses
Difference.
2.14) relative prediction residual is judged whether within ± 5%, and step 2.15 is performed if meeting and requiring), otherwise
Return to step 2.12), re-start training.
2.15) output meets desired flue gas oxygen content forecast model.
Communication module.Satisfaction is required that flue gas oxygen content forecast model sends function Distributed Control System, life to by the module
Produce management system.
A kind of circulating fluid bed domestic garbage burning boiler furnace outlet flue gas oxygen content Forecasting Methodology, this method include with
Lower step:
1) operation mechanism and flue gas oxygen content change mechanism of circulating fluid bed domestic garbage burning boiler are analyzed, rubbish is selected
Feeding coal, coal-supplying amount, primary air flow, secondary air flow, bed temperature, burner hearth freeboard temperature and the combustion chamber draft of rubbish are used as flue gas
The input variable of oxygen content forecast model.
2) training sample is gathered.The historical data that input variable is gathered from database is spaced according to set time, or
The operational factor under operating mode is specified in collection, constitutes the training sample matrix X (m × n) of flue gas oxygen content forecast model input variable,
M represents number of samples, and n represents the number of variable, is trained while gathering corresponding flue gas oxygen content as the output of model
Sample Y (m × 1);
3) data prediction.Gross error processing and random crror processing are carried out to X (m × n), those are not to forgo
Reflect the deceptive information of boiler accidental conditions, the unusual service conditions such as boiler shutdown, banking fire, batcher blocking are excluded, are
The harmful effects that dimension and the difference of the order of magnitude are caused to model performance between the parameter of forecast model are avoided, training sample is defeated
Enter variable to be mapped to after normalized in [0,1] interval, the training sample X of the input variable after being standardized*
The training sample Y of (m × n) and output variable*(m×1)。
4) intelligent algorithm integrated moulding.First with punishment parameter C and core of the Genetic Particle Swarm Algorithm on multiple populations to SVM models
Parameter g carries out optimizing, obtained best parameter group then is assigned into SVM models, and be trained based on this.Algorithm steps
It is rapid as follows:
4.1) initialization algorithm parameter.The maximum optimizing algebraically T of particle cluster algorithm on multiple populationsmax, maximum inertia weight
ωmax, minimum inertia weight ωmin, speed renewal coefficients R1、R2、R3、R4, population quantity pop, the number of particles of single species
Ind, migration rate Pi, crossover probability Pc, mutation probability Pm, hereditary generation gap Pe, genetic manipulation frequency Pg。
4.2) population is initialized.By the way of real coding, the punishment parameter C and nuclear parameter g of SVM models are compiled in order
Each weight threshold is generated as a real number between [0,200] by code at random among a particle.
4.3) individual extreme value and colony's extreme value are initialized.The parameter combination included in each particle is assigned to SVM models, and
Combined training sample carries out learning training, and the forecast model obtained using training calculates flue gas oxygen content predicted valueBy predicted valueWith actual measured value y*It is compared, and the fitness value fitness using error sum of squares MSE as particle, fitness calculating
Formula is as follows:
Initialization extreme value of the fitness value that the calculating of each particle is obtained as the particle in itself, each population it is central
Value minimum MSE is used as colony's extreme value.
4.4) more new particle.According to newest individual extreme value and global extremum, according to (2) formula and (3) formula more new particle
Speed vid(t) with position xid(t):
vid(t+1)=ω vid(t)+c1r1(pid-xid(t))+c2r2(pgd-xid(t)) (2)
xid(t+1)=xid(t)+vid(t+1) (3)
In formula, t is the optimizing algebraically of particle swarm optimization algorithm, r1, r2It is the random number between [0,1], pidRefer to
I-th of particle is sought so far searches optimal location, pgdRefer to the population and search for optimal location so far.Further, it is
Improve basic particle group algorithm and be easily trapped into local extremum and the slow defect of convergence rate, introduced on the basis of PSO algorithms
Dynamic aceleration pulse c1、c2With inertia weight ω:
Wherein, TmaxFor maximum optimizing algebraically, ωmaxFor maximum inertia weight, ωminFor minimum inertia weight, R1、R2、R3、
R4For constant.
4.5) judge whether to need to carry out genetic manipulation.According to genetic manipulation frequency PgCarry out, generally per PgIn generation, is carried out once
Genetic manipulation.Step 4.6 is then performed to carry out genetic manipulation), otherwise perform step 4.7)
4.6) genetic manipulation is performed.Genetic manipulation includes selecting, intersecting, making a variation, staying excellent, specific as follows.Selection:As wheel disc
Gambling equally calculates select probability, and equidistant selection is individual, if N is to need the individual amount of selection, the distance of select finger is
1/N, the random number of the position of first select finger in [0,1/N] interval is determined;Intersect:The method recombinated using arithmetic,
I.e. two individuals produce new individual by linear combination.Assuming that in two individualsWithBetween carry out arithmetic crossover, then press
Formula (7) and (8) obtain two new individualsWith
λ in formula1And λ2For the real number between (0,1), and λ1+λ2=1;Variation:Uniformity is taken to make a variation, it is assumed that s=(v1,
v2,...,vn) it is parent, z=(z1,z2,...,zn) it is the offspring that variation is produced.One is randomly chosen in present parent vector
Component, it is assumed that be k-th, then in its interval of definition [ak,bk] in take a number v' uniformly randomlykInstead of vkTo obtain zi;
Stay excellent:Using the strategy for retaining parent elite individual, according to the target function value of parent chromosome, replaced with transition for chromosome
Change in parent P after target function value rankinge× 100% chromosome, PeFor replacement rate (hereditary generation gap), in this way, part in parent
Optimal individual is retained, and will not be disappeared because of computings such as selection, intersection, variations.
4.7) particle fitness value calculation.The fitness value of particle after updating is calculated according to formula (1).
4.8) more new individual extreme value and colony's extreme value.Using fitness value as evaluation index, relatively more contemporary particle and previous generation
Fitness value size between particle, if the fitness value of current particle is better than previous generation, the position of current particle is set
Individual extreme value is set to, otherwise individual extreme value keeps constant.While the optimal particle of contemporary all particle fitness values is obtained, and with
Previous generation optimal particles are compared, if the fitness value of contemporary optimal particle is better than the fitness of previous generation optimal particles
Value, then be set to global optimum by the adaptive optimal control angle value of contemporary particle, and otherwise global optimum keeps constant.
4.9) immigrant's operation.In nature, the different populations of a species is distributed in different regions, on the one hand each
Groupy phase has facilitated their different existence to independently striving for resource from nature used in oneself under different regional conditions
Pattern and evolution degree;On the other hand, possibly through migrating between each population, make to contact each other, reach mutually
Be connected with without, learn from other's strong points to offset one's weaknesses, the purpose of common evolutionary.(Multi-Population, MP) on multiple populations hereditary grain that this patent is used
Swarm optimization has exactly used for reference this phenomenon of generally existing in nature.Each time in searching process, using single between population
The excellent individual in immigrant's operation, the 1st population is carried out to the mode of circulation migration and moves to the 2nd, and the 2nd is moved to the 3rd, with
This analogizes, and to the last one is moved to first.Migration rate P between populationiP before ranking in=0.04, i.e. expression source populationi
P after ranking among × 100% individual replacement target populationi× 100% individual, optimal knowledge between population is completed with this
Exchange.
4.10) algorithm stop condition algorithm judges.Judge whether to reach maximum iteration or reach precision of prediction
It is required that, the return to step 4.4 if being not reaching to), continued search for using the particle of renewal, otherwise exit search, perform step
4.11)。
4.11) optimal parameter combination particle is exported.
4.12) parameter combination among optimal particle is assigned to SVM models, and combined training sample is learnt.
4.13) precision of prediction of model is verified.The predicted value and actual value of model are contrasted, relative prediction is calculated and misses
Difference.
414) relative prediction residual is judged whether within ± 5%, and step 4.15 is performed if meeting and requiring), otherwise
Return to step 4.12), re-start training.
4.15) output meets desired flue gas oxygen content forecast model.
5) model adaptation updates.When the error of flue gas oxygen content and model predication value exceedes ± 5%, mould is updated immediately
Type.
The beneficial effects of the invention are as follows:Gone through using the operation mechanism and operation of circulating fluid bed domestic garbage burning boiler
In history data on the basis of tacit knowledge, using algorithm of support vector machine and genetic particle colony optimization algorithm on multiple populations is integrated builds
A kind of method of mould, the system and method for constructing fast, economical and adaptive updates are entered to boiler furnace outlet flue gas oxygen content
Row real-time estimate, avoids the modelling by mechanism work of very complicated.Wherein, the nonlinear dynamic characteristic of SVM algorithm, general is utilized
Change ability and real-time estimate ability characterize the dynamic variation characteristic of flue gas oxygen content, are that operations staff and designer grasp
The variation characteristic for solving flue gas oxygen content provides new approach;SVM algorithm punishment parameter C and core are joined using particle swarm optimization algorithm
Number g is optimized, and improves the generalization ability of model;Migration mechanism on multiple populations is introduced, the various of particle swarm optimization algorithm solution is improved
Property, reduce particle cluster algorithm optimizing and calculate the possibility for being absorbed in local optimum;Introduce the genetic manipulations such as selection, intersection, variation, fusion
The characteristic of genetic algorithm parallel search, can search multiple local best points, substantially increase ability of searching optimum and
Search speed, while historical information can be effectively utilized to speculate optimizing point set that expected performance of future generation increases;It is different
The computing environment of structure, substantially increases model construction efficiency, computer resource has been played the performance of maximum.Whole modeling process
Clear logic, auto-modeling degree is high, it is easy to grasps and promotes.Well-drilled flue gas oxygen content forecast model can be used for
The actual moving process of operations staff is instructed, those System design based on model algorithms can be serviced, or be used as soft measuring instrument
Check is complementary to one another with flue gas oxygen content hard ware measure system.
Brief description of the drawings
Fig. 1 is the structure chart of system proposed by the invention.
Fig. 2 is the structure chart of master system proposed by the invention.
Fig. 3 is the flow chart of intelligent modeling method proposed by the invention.
Embodiment
The present invention is described in further detail with specific embodiment below in conjunction with the accompanying drawings.
Embodiment 1
Reference picture 1, Fig. 2, Fig. 3, a kind of circulating fluid bed domestic garbage burning boiler furnace outlet cigarette that the present invention is provided
Gas oxygen content forecasting system, including circulating fluid bed domestic garbage burning boiler, and for the collecting and distributing of boiler operatiopn control
Control system, data communication interface, database and host computer.Database is by data communication interface from Distributed Control System
Data are read, and for the training study and test of host computer, host computer is entered by data communication interface with Distributed Control System
Row data exchange, described host computer includes on-line study, online updating, verification portion and on-line prediction part.Specifically include:
Signal acquisition module.The module is being burned when specifying house refuse for gathering CFB Domestic refuse incinerators
Operating condition state parameter and performance variable, and constitute refuse thermal value forecast model input variable training sample matrix X (m ×
N), m represents number of samples, and n represents the number of variable;
Data preprocessing module.Gross error processing and random crror processing are carried out to X (m × n), to forgo those not
It is the deceptive information for reflecting boiler accidental conditions, the unusual service conditions such as boiler shutdown, banking fire, batcher blocking is excluded,
In order to avoid the harmful effect that the difference of dimension between the parameter of forecast model and the order of magnitude is caused to model performance, training sample
Input variable is mapped to after normalized in [0,1] interval, the training sample X after being standardized*(m×n).In advance
Processing procedure is carried out using following steps:
1.1) criterion is reached according to Rye, rejects the outlier in training sample X (m × n);
1.2) boiler shutdown operating condition is rejected, the aperture of burner hearth feeder and batcher is zero during boiler shutdown, and
Temperature is close to normal temperature in burner hearth;
1.3) burner hearth banking fire operation conditions, primary air fan, overfire air fan air-introduced machine burner hearth feeder during boiler pressuring fire are rejected
Aperture with batcher is zero, but the temperature of burner hearth emulsion zone maintains 350 DEG C~450 DEG C;
1.4) reject batcher and block operating mode, batcher, which is blocked, needs what operations staff was shot by the camera of material inlet
Picture judges that feed situation when batcher is blocked, operations staff can significantly heighten the aperture of batcher, be reflected in fortune
In row data, i.e., the aperture of batcher is more than 35%;
1.5) data normalization is handled.According to formula (1) by the interval of data variable mappings to [0 1].
X in formulaJThe vector that J variables are constituted is represented, min () represents minimum value, and max () represents maximum.
Expertise library module.Training sample is constantly updated using the method for rolling time window so that training sample begins
Be maintained at eventually in newest state, rolling time window method refers to since current time, backtracking L (unit second) length when
Between size.
Model modification determination module.The performance of current predictive model is detected, when relative prediction residual exceedes ± 5%, then
Decision model needs to be updated.
Intelligent modeling module.Intelligent modeling module is the core of flue gas oxygen content forecasting system, the module first with
Genetic Particle Swarm Algorithm on multiple populations carries out optimizing to the punishment parameter C and nuclear parameter g of SVM models,
Then obtained best parameter group is assigned to SVM models, and be trained based on this.Algorithm steps are such as
Under:
2.1) initialization algorithm parameter.The maximum optimizing algebraically T of particle cluster algorithm on multiple populationsmax=100, maximum inertia power
Weight ωmax=1.4, minimum inertia weight ωmin=0.4, speed updates coefficients R1=1, R2=0.5, R3=6, R4=2, population number
Measure pop=5, the number of particles ind=20 of single species, migration rate Pi=0.04, crossover probability Pc=0.8, mutation probability Pm=
0.04th, hereditary generation gap Pe=0.05, genetic manipulation frequency Pg=2.
2.2) population is initialized.By the way of real coding, the punishment parameter C and nuclear parameter g of SVM models are compiled in order
Each weight threshold is generated as a real number between [0,200] by code at random among a particle.
2.3) individual extreme value and colony's extreme value are initialized.The parameter combination included in each particle is assigned to SVM models, and
Combined training sample carries out learning training, and SVM algorithm is described as follows:
Assuming that the training sample set of SVM models is (xi,yi), i=1 ..., n, x ∈ Rd, y ∈ R, mode input output variable
Between relation it is as follows:
In formula, using nonlinear functionSample data is mapped in a high-dimensional feature space (Hilbert);B is
Amount of bias.The key problem of SVM algorithm training is just to solve for ωTAnd b so that have for the input x beyond sample
In order to ensure that model has solution, it is contemplated that measurement data be difficult to avoid that containing noise, it is necessary to ideally
Linear regression function (2) is modified, that is, introduces the linear insensitive loss functions of ε and penalty factor and numerically more than zero
Slack variable ξi、Thus, solving convex quadratic programming problem turns into the key of solved function regression problem:
Lagrange multiplier is introduced, and the kernel function K (x for meeting Mercer conditions are constituted using mapping function φi, x),
According to KKT conditions, the Function Fitting equation that can be optimal
Wherein, xiFor supporting vector, nsv is the number of supporting vector.Generalization ability of the selection of kernel function to SVM models
There is more significant impact.From the higher Gaussian radial basis function of flexibility and versatility (RBF) as kernel function, it is expressed
Formula is:
K(xi, x)=exp (- g | | xi-x||2) (6)
In formula, g is nuclear parameter, g>0;||·||2For 2- norms.
The forecast model obtained using training calculates flue gas oxygen content predicted valueBy predicted valueWith actual measured value y*Enter
Row compares, and the fitness value fitness using error sum of squares MSE as particle, and fitness calculation formula is as follows:
Initialization extreme value of the fitness value that the calculating of each particle is obtained as the particle in itself, each population it is central
Value minimum MSE is used as colony's extreme value.
2.4) more new particle.According to newest individual extreme value and global extremum, according to (8) formula and (9) formula more new particle
Speed vid(t) with position xid(t):
vid(t+1)=ω vid(t)+c1r1(pid-xid(t))+c2r2(pgd-xid(t)) (8)
xid(t+1)=xid(t)+vid(t+1) (9)
In formula, t is the optimizing algebraically of particle swarm optimization algorithm, r1, r2It is the random number between [0,1], pidRefer to
I-th of particle is sought so far searches optimal location, pgdRefer to the population and search for optimal location so far.Further, it is
Improve basic particle group algorithm and be easily trapped into local extremum and the slow defect of convergence rate, introduced on the basis of PSO algorithms
Dynamic aceleration pulse c1、c2With inertia weight ω:
Wherein, TmaxFor maximum optimizing algebraically, ωmaxFor maximum inertia weight, ωminFor minimum inertia weight, R1、R2、R3、
R4For constant.
2.5) judge whether to need to carry out genetic manipulation.According to genetic manipulation frequency PgCarry out, generally per PgIn generation, is carried out once
Genetic manipulation.Step 2.6 is then performed to carry out genetic manipulation), otherwise perform step 2.7)
2.6) genetic manipulation is performed.Genetic manipulation includes selecting, intersecting, making a variation, staying excellent, specific as follows.Selection:As wheel disc
Gambling equally calculates select probability, and equidistant selection is individual, if N is to need the individual amount of selection, the distance of select finger is
1/N, the random number of the position of first select finger in [0,1/N] interval is determined;Intersect:The method recombinated using arithmetic,
I.e. two individuals produce new individual by linear combination.Assuming that in two individualsWithBetween carry out arithmetic crossover, then press
Formula (13) and (14) obtain two new individualsWith
λ in formula1And λ2For the real number between (0,1), and λ1+λ2=1;Variation:Uniformity is taken to make a variation, it is assumed that s=(v1,
v2,...,vn) it is parent, z=(z1,z2,...,zn) it is the offspring that variation is produced.One is randomly chosen in present parent vector
Component, it is assumed that be k-th, then in its interval of definition [ak,bk] in take a number v' uniformly randomlykInstead of vkTo obtain zi;
Stay excellent:Using the strategy for retaining parent elite individual, according to the target function value of parent chromosome, replaced with transition for chromosome
P after target function value ranking in parente× 100% chromosome, PeFor replacement rate (hereditary generation gap), in this way, part is most in parent
Excellent individual is retained, and will not be disappeared because of computings such as selection, intersection, variations.
2.7) particle fitness value calculation.The fitness value of particle after updating is calculated according to formula (7).
2.8) more new individual extreme value and colony's extreme value.Using fitness value as evaluation index, relatively more contemporary particle and previous generation
Fitness value size between particle, if the fitness value of current particle is better than previous generation, the position of current particle is set
Individual extreme value is set to, otherwise individual extreme value keeps constant.While the optimal particle of contemporary all particle fitness values is obtained, and with
Previous generation optimal particles are compared, if the fitness value of contemporary optimal particle is better than the fitness of previous generation optimal particles
Value, then be set to global optimum by the adaptive optimal control angle value of contemporary particle, and otherwise global optimum keeps constant.
2.9) immigrant's operation.In nature, the different populations of a species is distributed in different regions, on the one hand each
Groupy phase has facilitated their different existence to independently striving for resource from nature used in oneself under different regional conditions
Pattern and evolution degree;On the other hand, possibly through migrating between each population, make to contact each other, reach mutually
Be connected with without, learn from other's strong points to offset one's weaknesses, the purpose of common evolutionary.(Multi-Population, MP) on multiple populations hereditary grain that this patent is used
Swarm optimization has exactly used for reference this phenomenon of generally existing in nature.Each time in searching process, using single between population
The excellent individual in immigrant's operation, the 1st population is carried out to the mode of circulation migration and moves to the 2nd, and the 2nd is moved to the 3rd, with
This analogizes, and to the last one is moved to first.Migration rate P between populationiP before ranking in=0.04, i.e. expression source populationi
P after ranking among × 100% individual replacement target populationi× 100% individual, optimal knowledge between population is completed with this
Exchange.
2.10) algorithm stop condition algorithm judges.Judge whether to reach maximum iteration or reach precision of prediction
It is required that, the return to step 2.4 if being not reaching to), continued search for using the particle of renewal, otherwise exit search, perform step
2.11)。
2.11) optimal parameter combination particle is exported.
2.12) parameter combination among optimal particle is assigned to SVM models, and combined training sample is learnt.
2.13) precision of prediction of model is verified.The predicted value and actual value of model are contrasted, relative prediction is calculated and misses
Difference.
2.14) relative prediction residual is judged whether within ± 5%, and step 2.15 is performed if meeting and requiring), otherwise
Return to step 2.12), re-start training.
2.15) output meets desired flue gas oxygen content forecast model.
Communication module.Satisfaction is required that flue gas oxygen content forecast model sends function Distributed Control System, life to by the module
Produce management system.
Embodiment 2
Reference picture 1, Fig. 2, Fig. 3, a kind of circulating fluid bed domestic garbage burning boiler furnace outlet cigarette that the present invention is provided
Gas oxygen content Forecasting Methodology, this method comprises the following steps:
1) operation mechanism and flue gas oxygen content change mechanism of circulating fluid bed domestic garbage burning boiler are analyzed, rubbish is selected
Feeding coal, coal-supplying amount, primary air flow, secondary air flow, bed temperature, burner hearth freeboard temperature and the combustion chamber draft of rubbish are used as flue gas
The input variable of oxygen content forecast model.
Domestic domestic waste is generally mixed collection, is caused into factory, to enter stove component of refuse complex, general main
Including main components such as rubbish from cooking, paper, plastics, rubber, fabric, wood, bamboo and inorganic matters, low heat value, height are shown
Moisture and the larger feature of fluctuation.In order to ensure the stable burning of refuse incinerator of circulating fluid bed, it will usually add coal
It is used as auxiliary fuel.Burning of the rubbish in recirculating fluidized bed is a sufficiently complex vigorous physical chemical change process, rubbish
Rubbish can undergo several processes after burner hearth is entered:Dry heat, Volatile and burning, coke burning.Light weight in rubbish
Frangible component such as paper paper, plastics and fine grained can enter upper furnace in the presence of fluidized wind, and experience is dried, waved
The precipitation of hair point and a series of processes such as burning and the burning of carbon residue;And density is larger, moisture content is high and particle size compared with
The component that the big component such as terminal velocity such as wood, rubbish from cooking is more than fluidizing velocity can fall into emulsion zone, and in emulsion zone
Heated, burn by bed, different from the heat release rule of coal, the component of rubbish higher moisture low heat value can inhale in emulsion zone
Substantial amounts of heat is received, and substantial amounts of volatile matter burns in suspension section.
CFB furnace thorax exiting flue gas oxygen content is by the distribution of thermo parameters method situation and organic volatile matter concentration in burner hearth
What situation was determined.Coal-supplying amount, feeding coal and a secondary air flow have together decided on thermo parameters method, oxygen concentration distribution and organic waved
Hair point concentration distribution, they are reflected by bed temperature, burner hearth freeboard temperature and flue gas oxygen content.Especially note that
, in actual moving process, it may appear that the uneven situation of temperature field, component field distribution, and by Oxygen Amount in Flue Gas measuring point and
Fire box temperature measuring point can not be known completely, and the fluctuation situation of combustion chamber draft can reflect their fluctuation feelings to a certain extent
Condition, therefore also it regard it as one of input variable of model.
2) training sample is gathered.The historical data that input variable is gathered from database is spaced according to set time, or
The operational factor under operating mode is specified in collection, constitutes the training sample matrix X (m × n) of flue gas oxygen content forecast model input variable,
M represents number of samples, and n represents the number of variable, is trained while gathering corresponding flue gas oxygen content as the output of model
Sample Y (m × 1);
3) data prediction.Gross error processing and random crror processing are carried out to X (m × n), those are not to forgo
Reflect the deceptive information of boiler accidental conditions, the unusual service conditions such as boiler shutdown, banking fire, batcher blocking are excluded, are
The harmful effects that dimension and the difference of the order of magnitude are caused to model performance between the parameter of forecast model are avoided, training sample is defeated
Enter variable to be mapped to after normalized in [0,1] interval, the training sample X of the input variable after being standardized*
The training sample Y of (m × n) and output variable*(m×1)。
4) intelligent algorithm integrated moulding.First with punishment parameter C and core of the Genetic Particle Swarm Algorithm on multiple populations to SVM models
Parameter g carries out optimizing, obtained best parameter group then is assigned into SVM models, and be trained based on this.Algorithm steps
It is rapid as follows:
4.1) initialization algorithm parameter.The maximum optimizing algebraically T of particle cluster algorithm on multiple populationsmax=100, maximum inertia power
Weight ωmax=1.4, minimum inertia weight ωmin=0.4, speed updates coefficients R1=1, R2=0.5, R3=6, R4=2, population number
Measure pop=5, the number of particles ind=20 of single species, migration rate Pi=0.04, crossover probability Pc=0.8, mutation probability Pm=
0.04th, hereditary generation gap Pe=0.05, genetic manipulation frequency Pg=2.
4.2) population is initialized.By the way of real coding, the punishment parameter C and nuclear parameter g of SVM models are compiled in order
Each weight threshold is generated as a real number between [0,200] by code at random among a particle.
4.3) individual extreme value and colony's extreme value are initialized.The parameter combination included in each particle is assigned to SVM models, and
Combined training sample carries out learning training, and SVM algorithm is described as follows:
Assuming that the training sample set of SVM models is (xi,yi), i=1 ..., n, x ∈ Rd, y ∈ R, mode input output variable
Between relation it is as follows:
In formula, using nonlinear functionSample data is mapped in a high-dimensional feature space (Hilbert);B is
Amount of bias.The key problem of SVM algorithm training is just to solve for ωTAnd b so that have for the input x beyond sample
In order to ensure that model has solution, it is contemplated that measurement data be difficult to avoid that containing noise, it is necessary to ideally
Linear regression function (1) is modified, that is, introduces the linear insensitive loss functions of ε and penalty factor and numerically more than zero
Slack variable ξi、Thus, solving convex quadratic programming problem turns into the key of solved function regression problem:
Lagrange multiplier is introduced, and the kernel function K (x for meeting Mercer conditions are constituted using mapping function φi, x),
According to KKT conditions, the Function Fitting equation that can be optimal
Wherein, xiFor supporting vector, nsv is the number of supporting vector.Generalization ability of the selection of kernel function to SVM models
There is more significant impact.From the higher Gaussian radial basis function of flexibility and versatility (RBF) as kernel function, it is expressed
Formula is:
K(xi, x)=exp (- g | | xi-x||2) (5)
In formula, g is nuclear parameter, g>0;||·||2For 2- norms.
The forecast model obtained using training calculates flue gas oxygen content predicted valueBy predicted valueWith actual measured value y*Enter
Row compares, and the fitness value fitness using error sum of squares MSE as particle, and fitness calculation formula is as follows:
Initialization extreme value of the fitness value that the calculating of each particle is obtained as the particle in itself, each population it is central
Value minimum MSE is used as colony's extreme value.
4.4) more new particle.According to newest individual extreme value and global extremum, according to (7) formula and (8) formula more new particle
Speed vid(t) with position xid(t):
vid(t+1)=ω vid(t)+c1r1(pid-xid(t))+c2r2(pgd-xid(t)) (7)
xid(t+1)=xid(t)+vid(t+1) (8)
In formula, t is the optimizing algebraically of particle swarm optimization algorithm, r1, r2It is the random number between [0,1], pidRefer to
I-th of particle is sought so far searches optimal location, pgdRefer to the population and search for optimal location so far.Further, it is
Improve basic particle group algorithm and be easily trapped into local extremum and the slow defect of convergence rate, introduced on the basis of PSO algorithms
Dynamic aceleration pulse c1、c2With inertia weight ω:
Wherein, TmaxFor maximum optimizing algebraically, ωmaxFor maximum inertia weight, ωminFor minimum inertia weight, R1、R2、R3、
R4For constant.
4.5) judge whether to need to carry out genetic manipulation.According to genetic manipulation frequency PgCarry out, generally per PgIn generation, is carried out once
Genetic manipulation.Step 4.6 is then performed to carry out genetic manipulation), otherwise perform step 4.7)
4.6) genetic manipulation is performed.Genetic manipulation includes selecting, intersecting, making a variation, staying excellent, specific as follows.Selection:As wheel disc
Gambling equally calculates select probability, and equidistant selection is individual, if N is to need the individual amount of selection, the distance of select finger is
1/N, the random number of the position of first select finger in [0,1/N] interval is determined;Intersect:The method recombinated using arithmetic,
I.e. two individuals produce new individual by linear combination.Assuming that in two individualsWithBetween carry out arithmetic crossover, then press
Formula (12) and (13) obtain two new individualsWith
λ in formula1And λ2For the real number between (0,1), and λ1+λ2=1;Variation:Uniformity is taken to make a variation, it is assumed that s=(v1,
v2,...,vn) it is parent, z=(z1,z2,...,zn) it is the offspring that variation is produced.One is randomly chosen in present parent vector
Component, it is assumed that be k-th, then in its interval of definition [ak,bk] in take a number v' uniformly randomlykInstead of vkTo obtain zi;
Stay excellent:Using the strategy for retaining parent elite individual, according to the target function value of parent chromosome, replaced with transition for chromosome
P after target function value ranking in parente× 100% chromosome, PeFor replacement rate (hereditary generation gap), in this way, part is most in parent
Excellent individual is retained, and will not be disappeared because of computings such as selection, intersection, variations.
4.7) particle fitness value calculation.The fitness value of particle after updating is calculated according to formula (6).
4.8) more new individual extreme value and colony's extreme value.Using fitness value as evaluation index, relatively more contemporary particle and previous generation
Fitness value size between particle, if the fitness value of current particle is better than previous generation, the position of current particle is set
Individual extreme value is set to, otherwise individual extreme value keeps constant.While the optimal particle of contemporary all particle fitness values is obtained, and with
Previous generation optimal particles are compared, if the fitness value of contemporary optimal particle is better than the fitness of previous generation optimal particles
Value, then be set to global optimum by the adaptive optimal control angle value of contemporary particle, and otherwise global optimum keeps constant.
4.9) immigrant's operation.In nature, the different populations of a species is distributed in different regions, on the one hand each
Groupy phase has facilitated their different existence to independently striving for resource from nature used in oneself under different regional conditions
Pattern and evolution degree;On the other hand, possibly through migrating between each population, make to contact each other, reach mutually
Be connected with without, learn from other's strong points to offset one's weaknesses, the purpose of common evolutionary.(Multi-Population, MP) on multiple populations hereditary grain that this patent is used
Swarm optimization has exactly used for reference this phenomenon of generally existing in nature.Each time in searching process, using single between population
The excellent individual in immigrant's operation, the 1st population is carried out to the mode of circulation migration and moves to the 2nd, and the 2nd is moved to the 3rd, with
This analogizes, and to the last one is moved to first.Migration rate P between populationiP before ranking in=0.04, i.e. expression source populationi
P after ranking among × 100% individual replacement target populationi× 100% individual, optimal knowledge between population is completed with this
Exchange.
4.10) algorithm stop condition algorithm judges.Judge whether to reach maximum iteration or reach precision of prediction
It is required that, the return to step 4.4 if being not reaching to), continued search for using the particle of renewal, otherwise exit search, perform step
4.11)。
4.11) optimal parameter combination particle is exported.
4.12) parameter combination among optimal particle is assigned to SVM models, and combined training sample is learnt.
4.13) precision of prediction of model is verified.The predicted value and actual value of model are contrasted, relative prediction is calculated and misses
Difference.
414) relative prediction residual is judged whether within ± 5%, and step 4.15 is performed if meeting and requiring), otherwise
Return to step 4.12), re-start training.
4.15) output meets desired flue gas oxygen content forecast model.
5) model adaptation updates.When the error of flue gas oxygen content and model predication value exceedes ± 5%, mould is updated immediately
Type.
Claims (2)
1. a kind of real-time estimate system of circulating fluid bed domestic garbage burning boiler furnace outlet flue gas oxygen content, its feature exists
In the system is connected with the Distributed Control System and production management system of CFBB, including data communication interface
And host computer, flue gas oxygen content forecast model is trained and updated in host computer, then passes through the model trained
Communication interface is sent to Distributed Control System, production management system, and the host computer includes:
Signal acquisition module.The module is used to gather operation of the CFB Domestic refuse incinerators when burning specified house refuse
Work condition state parameter and performance variable, and constitute the training sample matrix X (m × n) of refuse thermal value forecast model input variable, m
Number of samples is represented, n represents the number of variable.
Data preprocessing module.Gross error processing and random crror processing are carried out to X (m × n), the open country in training sample is rejected
Value, excludes the unusual service conditions such as boiler shutdown, banking fire, batcher blocking, and training sample input variable is reflected after normalized
It is mapped in [0,1] interval, the training sample X after being standardized*(m×n)。
Expertise library module.Training sample is constantly updated using the method for rolling time window so that training sample is protected all the time
Hold in newest state, rolling time window method refers to since current time, the time chi of backtracking L (unit second) length
It is very little.
Model modification determination module.The performance of current predictive model is detected, when relative prediction residual exceedes ± 5%, is then judged
Model needs to be updated.
Intelligent modeling module.The module is first with punishment parameter C and nuclear parameter of the Genetic Particle Swarm Algorithm on multiple populations to SVM models
G carries out optimizing, obtained best parameter group then is assigned into SVM models, and be trained based on this.Specific steps are such as
Under:
2.1) initialization algorithm parameter.The maximum optimizing algebraically T of particle cluster algorithm on multiple populationsmax, maximum inertia weight ωmax, most
Small inertia weight ωmin, speed renewal coefficients R1、R2、R3、R4, population quantity pop, the number of particles ind of single species, immigrant
Rate Pi, crossover probability Pc, mutation probability Pm, hereditary generation gap Pe, genetic manipulation frequency Pg。
2.2) population is initialized.By the way of real coding, by the punishment parameter C and nuclear parameter g of SVM models, coding exists in order
Among one particle, and each weight threshold is generated as to a real number between [0,200] at random.
2.3) individual extreme value and colony's extreme value are initialized.The parameter combination included in each particle is assigned to SVM models, and combined
Training sample carries out learning training, and the forecast model obtained using training calculates flue gas oxygen content predicted valueBy predicted valueWith
Actual measured value y*It is compared, and the fitness value fitness using error sum of squares MSE as particle, fitness calculates public
Formula is as follows:
Initialization extreme value of the fitness value that the calculating of each particle is obtained as the particle in itself, the central MSE of each population
Minimum value is used as colony's extreme value.
2.4) more new particle.According to newest individual extreme value and global extremum, according to (3) formula and the speed of (4) formula more new particle
vid(t) with position xid(t):
vid(t+1)=ω vid(t)+c1r1(pid-xid(t))+c2r2(pgd-xid(t)) (3)
xid(t+1)=xid(t)+vid(t+1) (4)
In formula, t is the optimizing algebraically of particle swarm optimization algorithm, r1, r2It is the random number between [0,1], pidRefer to i-th
Particle is sought so far searches optimal location, pgdRefer to the population and search for optimal location so far.Further, in order to improve
Basic particle group algorithm is easily trapped into local extremum and the slow defect of convergence rate, and dynamic has been introduced on the basis of PSO algorithms
Aceleration pulse c1、c2With inertia weight ω:
Wherein, TmaxFor maximum optimizing algebraically, ωmaxFor maximum inertia weight, ωminFor minimum inertia weight, R1、R2、R3、R4For
Constant.
2.5) judge whether to need to carry out genetic manipulation.According to genetic manipulation frequency PgCarry out, generally per PgIn generation, carries out once hereditary
Operation.Step 2.6 is then performed to carry out genetic manipulation), otherwise perform step 2.7)
2.6) genetic manipulation is performed.Genetic manipulation includes selecting, intersecting, making a variation, staying excellent, specific as follows.Selection:As roulette one
Sample calculates select probability, and equidistant selection is individual, if N is to need the individual amount of selection, the distance of select finger is 1/N,
Random number of the position of first select finger in [0,1/N] interval is determined;Intersect:The method recombinated using arithmetic, i.e., two
Individual produces new individual by linear combination.Assuming that in two individualsWithBetween carry out arithmetic crossover, then by formula
And (9) obtain two new individuals (8)With
λ in formula1And λ2For the real number between (0,1), and λ1+λ2=1;Variation:Uniformity is taken to make a variation, it is assumed that s=(v1,
v2,...,vn) it is parent, z=(z1,z2,...,zn) it is the offspring that variation is produced.One is randomly chosen in present parent vector
Component, it is assumed that be k-th, then in its interval of definition [ak,bk] in take a number v' uniformly randomlykInstead of vkTo obtain zk;
Stay excellent:Using the strategy for retaining parent elite individual, according to the target function value of parent chromosome, replaced with transition for chromosome
P after target function value ranking in parente× 100% chromosome, PeFor replacement rate (hereditary generation gap), in this way, part is most in parent
Excellent individual is retained, and will not be disappeared because of computings such as selection, intersection, variations.
2.7) particle fitness value calculation.The fitness value of particle after updating is calculated according to formula (2).
2.8) more new individual extreme value and colony's extreme value.Using fitness value as evaluation index, relatively more contemporary particle and previous generation particles
Between fitness value size, if the fitness value of current particle be better than previous generation, the position of current particle is set to
Individual extreme value, otherwise individual extreme value holding is constant.Obtain the optimal particle of contemporary all particle fitness values simultaneously, and with upper one
It is compared for optimal particle, if the fitness value of contemporary optimal particle is better than the fitness value of previous generation optimal particles,
The adaptive optimal control angle value of contemporary particle is set to global optimum, otherwise global optimum keeps constant.
2.9) immigrant's operation.Using Genetic Particle Swarm Algorithm on multiple populations, use and unidirectionally follow each time in searching process, between population
The excellent individual that the mode of ring migration is carried out in immigrant's operation, the 1st population moves to the 2nd, and the 2nd is moved to the 3rd, with such
Push away, to the last one is moved to first.Migration rate P between populationiP before ranking in=0.04, i.e. expression source populationi×
P after ranking among 100% individual replacement target populationi× 100% individual, the friendship of optimal knowledge between population is completed with this
Stream.
2.10) algorithm stop condition algorithm judges.Judge whether to reach maximum iteration or reach the requirement of precision of prediction,
The return to step 2.4 if being not reaching to), continued search for using the particle of renewal, otherwise exit search, perform step 2.11).
2.11) optimal parameter combination particle is exported.
2.12) parameter combination among optimal particle is assigned to SVM models, and combined training sample is learnt.
2.13) precision of prediction of model is verified.The predicted value and actual value of model are contrasted, relative prediction residual is calculated.
2.14) relative prediction residual is judged whether within ± 5%, and step 2.15 is performed if meeting and requiring), otherwise return
Step 2.12), re-start training.
2.15) output meets desired flue gas oxygen content forecast model.
Communication module.Satisfaction is required that flue gas oxygen content forecast model sends function Distributed Control System, production pipe to by the module
Reason system.
2. a kind of circulating fluid bed domestic garbage burning boiler furnace outlet flue gas oxygen content Forecasting Methodology, it is characterised in that should
Method comprises the following steps:
1) operation mechanism and flue gas oxygen content change mechanism of circulating fluid bed domestic garbage burning boiler are analyzed, selection rubbish
Feeding coal, coal-supplying amount, primary air flow, secondary air flow, bed temperature, burner hearth freeboard temperature and combustion chamber draft are oxygen-containing as flue gas
Measure the input variable of forecast model.
2) training sample is gathered.The historical data that input variable is gathered from database, or collection are spaced according to set time
The operational factor under operating mode is specified, the training sample matrix X (m × n) of flue gas oxygen content forecast model input variable, m tables is constituted
Show number of samples, n represents the number of variable, while gathering corresponding flue gas oxygen content as the output training sample of model
Y(m×1);
3) data prediction.Gross error processing and random crror processing are carried out to X (m × n), those are not reflection to forgo
The deceptive information of boiler accidental conditions, the unusual service conditions such as boiler shutdown, banking fire, batcher blocking is excluded, in order to keep away
Exempt from the harmful effects that dimension and the difference of the order of magnitude are caused to model performance between the parameter of forecast model, training sample input becomes
Amount is mapped to after normalized in [0,1] interval, the training sample X of the input variable after being standardized*(m×
N) with the training sample Y of output variable*(m×1)。
4) intelligent algorithm integrated moulding.First with punishment parameter C and nuclear parameter of the Genetic Particle Swarm Algorithm on multiple populations to SVM models
G carries out optimizing, obtained best parameter group then is assigned into SVM models, and be trained based on this.Algorithm steps are such as
Under:
4.1) initialization algorithm parameter.The maximum optimizing algebraically T of particle cluster algorithm on multiple populationsmax, maximum inertia weight ωmax, most
Small inertia weight ωmin, speed renewal coefficients R1、R2、R3、R4, population quantity pop, the number of particles ind of single species, immigrant
Rate Pi, crossover probability Pc, mutation probability Pm, hereditary generation gap Pe, genetic manipulation frequency Pg。
4.2) population is initialized.By the way of real coding, by the punishment parameter C and nuclear parameter g of SVM models, coding exists in order
Among one particle, and each weight threshold is generated as to a real number between [0,200] at random.
4.3) individual extreme value and colony's extreme value are initialized.The parameter combination included in each particle is assigned to SVM models, and combined
Training sample carries out learning training, and the forecast model obtained using training calculates flue gas oxygen content predicted valueBy predicted valueWith
Actual measured value y*It is compared, and the fitness value fitness using error sum of squares MSE as particle, fitness calculates public
Formula is as follows:
Initialization extreme value of the fitness value that the calculating of each particle is obtained as the particle in itself, the central MSE of each population
Minimum value is used as colony's extreme value.
4.4) more new particle.According to newest individual extreme value and global extremum, according to (2) formula and the speed of (3) formula more new particle
vid(t) with position xid(t):
vid(t+1)=ω vid(t)+c1r1(pid-xid(t))+c2r2(pgd-xid(t)) (2)
xid(t+1)=xid(t)+vid(t+1) (3)
In formula, t is the optimizing algebraically of particle swarm optimization algorithm, r1, r2It is the random number between [0,1], pidRefer to i-th
Particle is sought so far searches optimal location, pgdRefer to the population and search for optimal location so far.Further, in order to improve
Basic particle group algorithm is easily trapped into local extremum and the slow defect of convergence rate, and dynamic has been introduced on the basis of PSO algorithms
Aceleration pulse c1、c2With inertia weight ω:
Wherein, TmaxFor maximum optimizing algebraically, ωmaxFor maximum inertia weight, ωminFor minimum inertia weight, R1、R2、R3、R4For
Constant.
4.5) judge whether to need to carry out genetic manipulation.According to genetic manipulation frequency PgCarry out, generally per PgIn generation, carries out once hereditary
Operation.Step 4.6 is then performed to carry out genetic manipulation), otherwise perform step 4.7)
4.6) genetic manipulation is performed.Genetic manipulation includes selecting, intersecting, making a variation, staying excellent, specific as follows.Selection:As roulette one
Sample calculates select probability, and equidistant selection is individual, if N is to need the individual amount of selection, the distance of select finger is 1/N,
Random number of the position of first select finger in [0,1/N] interval is determined;Intersect:The method recombinated using arithmetic, i.e., two
Individual produces new individual by linear combination.Assuming that in two individualsWithBetween carry out arithmetic crossover, then by formula
And (8) obtain two new individuals (7)With
λ in formula1And λ2For the real number between (0,1), and λ1+λ2=1;Variation:Uniformity is taken to make a variation, it is assumed that s=(v1,
v2,...,vn) it is parent, z=(z1,z2,...,zn) it is the offspring that variation is produced.One is randomly chosen in present parent vector
Component, it is assumed that be k-th, then in its interval of definition [ak,bk] in take a number v' uniformly randomlykInstead of vkTo obtain zi;
Stay excellent:Using the strategy for retaining parent elite individual, according to the target function value of parent chromosome, replaced with transition for chromosome
Change in parent P after target function value rankinge× 100% chromosome, PeFor replacement rate (hereditary generation gap), in this way, part in parent
Optimal individual is retained, and will not be disappeared because of computings such as selection, intersection, variations.
4.7) particle fitness value calculation.The fitness value of particle after updating is calculated according to formula (1).
4.8) more new individual extreme value and colony's extreme value.Using fitness value as evaluation index, relatively more contemporary particle and previous generation particles
Between fitness value size, if the fitness value of current particle be better than previous generation, the position of current particle is set to
Individual extreme value, otherwise individual extreme value holding is constant.Obtain the optimal particle of contemporary all particle fitness values simultaneously, and with upper one
It is compared for optimal particle, if the fitness value of contemporary optimal particle is better than the fitness value of previous generation optimal particles,
The adaptive optimal control angle value of contemporary particle is set to global optimum, otherwise global optimum keeps constant.
4.9) immigrant's operation.The Genetic Particle Swarm Algorithm on multiple populations used, each time in searching process, using unidirectional between population
The excellent individual that the mode of circulation migration is carried out in immigrant's operation, the 1st population moves to the 2nd, and the 2nd is moved to the 3rd, with this
Analogize, to the last one is moved to first.Migration rate P between populationiP before ranking in=0.04, i.e. expression source populationi×
P after ranking among 100% individual replacement target populationi× 100% individual, the friendship of optimal knowledge between population is completed with this
Stream.
4.10) algorithm stop condition algorithm judges.Judge whether to reach maximum iteration or reach the requirement of precision of prediction,
The return to step 4.4 if being not reaching to), continued search for using the particle of renewal, otherwise exit search, perform step 4.11).
4.11) optimal parameter combination particle is exported.
4.12) parameter combination among optimal particle is assigned to SVM models, and combined training sample is learnt.
4.13) precision of prediction of model is verified.The predicted value and actual value of model are contrasted, relative prediction residual is calculated.
414) relative prediction residual is judged whether within ± 5%, and step 4.15 is performed if meeting and requiring), otherwise return
Step 4.12), re-start training.
4.15) export the flue gas oxygen content forecast model for meeting and requiring.
5) model adaptation updates.When the error of flue gas oxygen content and model predication value exceedes ± 5%, more new model immediately.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710108399.9A CN107016455B (en) | 2017-02-27 | 2017-02-27 | Prediction system and method for oxygen content of flue gas at hearth outlet of circulating fluidized bed household garbage incineration boiler |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710108399.9A CN107016455B (en) | 2017-02-27 | 2017-02-27 | Prediction system and method for oxygen content of flue gas at hearth outlet of circulating fluidized bed household garbage incineration boiler |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107016455A true CN107016455A (en) | 2017-08-04 |
CN107016455B CN107016455B (en) | 2020-04-10 |
Family
ID=59439808
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710108399.9A Active CN107016455B (en) | 2017-02-27 | 2017-02-27 | Prediction system and method for oxygen content of flue gas at hearth outlet of circulating fluidized bed household garbage incineration boiler |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107016455B (en) |
Cited By (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108629495A (en) * | 2018-04-23 | 2018-10-09 | 光大环保技术研究院(深圳)有限公司 | A kind of evaluation method of refuse thermal value |
CN108809871A (en) * | 2018-05-08 | 2018-11-13 | 中天宽带技术有限公司 | A kind of equal power suppressing method in peaks OFDM based on Multiple-population Genetic Algorithm |
CN109034523A (en) * | 2018-06-11 | 2018-12-18 | 湖南工业大学 | A kind of Rotary Kiln operating condition operation mode optimization method of fusion least square method supporting vector machine recurrence and particle group optimizing |
CN109977559A (en) * | 2019-03-28 | 2019-07-05 | 东北大学 | A kind of Pulverized Fuel Boiler Furnace inverse radiation analysis method for solving based on improved adaptive GA-IAGA |
CN110334839A (en) * | 2019-04-15 | 2019-10-15 | 北京航空航天大学 | Flight is delayed prediction technique, device, equipment and storage medium |
CN111538355A (en) * | 2020-05-06 | 2020-08-14 | 安徽工业大学 | GA-IGPC-based boiler flue GAs oxygen content control method and system |
CN111651847A (en) * | 2019-02-18 | 2020-09-11 | 新奥数能科技有限公司 | Method and device for determining running oxygen amount of gas-fired boiler |
CN112036100A (en) * | 2020-09-24 | 2020-12-04 | 哈尔滨锅炉厂有限责任公司 | Method for predicting hearth oxygen concentration by using regression algorithm based on numerical simulation |
CN112308311A (en) * | 2020-10-29 | 2021-02-02 | 西安工业大学 | Online prediction system and method for oxygen content of flue gas of thermal power plant |
CN113054881A (en) * | 2021-02-22 | 2021-06-29 | 江苏大学 | Method for constructing outer rotor stator coreless bearingless permanent magnet synchronous motor controller |
CN113063916A (en) * | 2021-03-11 | 2021-07-02 | 武汉钢铁有限公司 | Hot-dip galvanized strip steel coating aluminum content prediction method based on PSO-SVR model |
CN113419050A (en) * | 2021-07-20 | 2021-09-21 | 山东恒拓科技发展有限公司 | Method and device for soft measurement of cement raw material components |
CN113610260A (en) * | 2021-04-24 | 2021-11-05 | 北京工业大学 | Method for predicting concentration of flue gas components in incineration process of municipal domestic waste |
CN114637212A (en) * | 2022-05-17 | 2022-06-17 | 浙江浩普智能科技有限公司 | Oxygen content prediction control method and system |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100241249A1 (en) * | 2006-04-25 | 2010-09-23 | Pegasus Technologies, Inc. | System for optimizing oxygen in a boiler |
CN103823971A (en) * | 2014-02-19 | 2014-05-28 | 上海交通大学 | Oxygen content detecting method for boiler fuel gas of coal-fired 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 |
-
2017
- 2017-02-27 CN CN201710108399.9A patent/CN107016455B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100241249A1 (en) * | 2006-04-25 | 2010-09-23 | Pegasus Technologies, Inc. | System for optimizing oxygen in a boiler |
CN103823971A (en) * | 2014-02-19 | 2014-05-28 | 上海交通大学 | Oxygen content detecting method for boiler fuel gas of coal-fired 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 |
Non-Patent Citations (2)
Title |
---|
CHANGLIANG LIU 等: "Soft-Sensing of Oxygen-content in Flue Gases of Power Plant Based on LS-SVM and Simplex Algorithm", 《2009 WRI WORLD CONGRESS ON SOFTWARE ENGINEERING》 * |
杨秀 等: "基于PSO-SVM的电厂锅炉排烟氧含量软测量研究", 《科技经济导刊》 * |
Cited By (20)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108629495A (en) * | 2018-04-23 | 2018-10-09 | 光大环保技术研究院(深圳)有限公司 | A kind of evaluation method of refuse thermal value |
CN108809871A (en) * | 2018-05-08 | 2018-11-13 | 中天宽带技术有限公司 | A kind of equal power suppressing method in peaks OFDM based on Multiple-population Genetic Algorithm |
CN109034523B (en) * | 2018-06-11 | 2022-06-28 | 湖南工业大学 | Volatilization kiln working condition operation mode optimization method integrating least square support vector machine regression and particle swarm optimization |
CN109034523A (en) * | 2018-06-11 | 2018-12-18 | 湖南工业大学 | A kind of Rotary Kiln operating condition operation mode optimization method of fusion least square method supporting vector machine recurrence and particle group optimizing |
CN111651847B (en) * | 2019-02-18 | 2023-11-03 | 新奥数能科技有限公司 | Method and device for determining operation oxygen amount of gas boiler |
CN111651847A (en) * | 2019-02-18 | 2020-09-11 | 新奥数能科技有限公司 | Method and device for determining running oxygen amount of gas-fired boiler |
CN109977559A (en) * | 2019-03-28 | 2019-07-05 | 东北大学 | A kind of Pulverized Fuel Boiler Furnace inverse radiation analysis method for solving based on improved adaptive GA-IAGA |
CN110334839A (en) * | 2019-04-15 | 2019-10-15 | 北京航空航天大学 | Flight is delayed prediction technique, device, equipment and storage medium |
US11501648B2 (en) | 2019-04-15 | 2022-11-15 | Beihang University | Method and apparatus for predicting flight delay, device and storage medium |
CN111538355A (en) * | 2020-05-06 | 2020-08-14 | 安徽工业大学 | GA-IGPC-based boiler flue GAs oxygen content control method and system |
CN111538355B (en) * | 2020-05-06 | 2023-02-24 | 安徽工业大学 | GA-IGPC-based boiler flue GAs oxygen content control method and system |
CN112036100A (en) * | 2020-09-24 | 2020-12-04 | 哈尔滨锅炉厂有限责任公司 | Method for predicting hearth oxygen concentration by using regression algorithm based on numerical simulation |
CN112308311A (en) * | 2020-10-29 | 2021-02-02 | 西安工业大学 | Online prediction system and method for oxygen content of flue gas of thermal power plant |
CN113054881B (en) * | 2021-02-22 | 2022-10-28 | 江苏大学 | Method for constructing outer rotor stator coreless bearingless permanent magnet synchronous motor controller |
CN113054881A (en) * | 2021-02-22 | 2021-06-29 | 江苏大学 | Method for constructing outer rotor stator coreless bearingless permanent magnet synchronous motor controller |
CN113063916A (en) * | 2021-03-11 | 2021-07-02 | 武汉钢铁有限公司 | Hot-dip galvanized strip steel coating aluminum content prediction method based on PSO-SVR model |
CN113610260A (en) * | 2021-04-24 | 2021-11-05 | 北京工业大学 | Method for predicting concentration of flue gas components in incineration process of municipal domestic waste |
CN113610260B (en) * | 2021-04-24 | 2024-03-29 | 北京工业大学 | Method for predicting concentration of smoke components in urban household garbage incineration process |
CN113419050A (en) * | 2021-07-20 | 2021-09-21 | 山东恒拓科技发展有限公司 | Method and device for soft measurement of cement raw material components |
CN114637212A (en) * | 2022-05-17 | 2022-06-17 | 浙江浩普智能科技有限公司 | Oxygen content prediction control method and system |
Also Published As
Publication number | Publication date |
---|---|
CN107016455B (en) | 2020-04-10 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107016455A (en) | The forecasting system and method for circulating fluid bed domestic garbage burning boiler furnace outlet flue gas oxygen content | |
CN106931453B (en) | The forecasting system and method for circulating fluid bed domestic garbage burning emission of NOx of boiler | |
CN106224939B (en) | Circulating fluid bed domestic garbage burning boiler bed temperature Forecasting Methodology and system | |
Shi et al. | Combustion optimization of ultra supercritical boiler based on artificial intelligence | |
CN105864797B (en) | Real-time prediction system and method for boiler entering heat value of circulating fluidized bed household garbage incineration boiler | |
CN107038334A (en) | Circulating fluid bed domestic garbage burning boiler CO emitted smoke system and methods | |
Chu et al. | Constrained optimization of combustion in a simulated coal-fired boiler using artificial neural network model and information analysis☆ | |
Zheng et al. | A comparative study of optimization algorithms for low NOx combustion modification at a coal-fired utility boiler | |
Han et al. | Resource optimization model using novel extreme learning machine with t-distributed stochastic neighbor embedding: Application to complex industrial processes | |
CN106055520B (en) | System and method for predicting drum water level of circulating fluidized bed domestic garbage incineration boiler | |
CN104763999A (en) | Power plant pulverized coal boiler combustion performance online optimizing method and system | |
Hou et al. | Coordinated control system modeling of ultra-supercritical unit based on a new fuzzy neural network | |
Sun et al. | Prediction of oxygen content using weighted PCA and improved LSTM network in MSWI process | |
Li et al. | Prediction of spontaneous combustion in the coal stockpile based on an improved metabolic grey model | |
CN109992844A (en) | A kind of boiler flyash carbon content prediction technique based on ADQPSO-SVR model | |
CN110991756A (en) | MSWI furnace temperature prediction method based on TS fuzzy neural network | |
Ma et al. | Three-objective optimization of boiler combustion process based on multi-objective teaching–learning based optimization algorithm and ameliorated extreme learning machine | |
Nemitallah et al. | Artificial intelligence for control and optimization of boilers’ performance and emissions: A review | |
Duan et al. | Prediction of NOx concentration using modular long short-term memory neural network for municipal solid waste incineration | |
Xu et al. | A new online optimization method for boiler combustion system based on the data-driven technique and the case-based reasoning principle | |
CN109140474A (en) | A kind of coal-fired power station boiler manually intelligent sootblowing | |
Movahed et al. | Modeling and optimization of NO emission for a steam power plant by data‐driven methods | |
Cui et al. | Multi-condition operational optimization with adaptive knowledge transfer for municipal solid waste incineration process | |
Li et al. | A combined NOx emission prediction model based on semi-empirical model and black box models | |
CN109034523A (en) | A kind of Rotary Kiln operating condition operation mode optimization method of fusion least square method supporting vector machine recurrence and particle group optimizing |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |